UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Energetics and cooling in urban parks Spronken-Smith, Rachel A. 1994

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1994-953950.pdf [ 5.92MB ]
JSON: 831-1.0088055.json
JSON-LD: 831-1.0088055-ld.json
RDF/XML (Pretty): 831-1.0088055-rdf.xml
RDF/JSON: 831-1.0088055-rdf.json
Turtle: 831-1.0088055-turtle.txt
N-Triples: 831-1.0088055-rdf-ntriples.txt
Original Record: 831-1.0088055-source.json
Full Text

Full Text

ENERGETICS AND COOLING IN URBAN PARKSByRachel A. Spronken—SmithB. Sc. (Hons.) University of Otago, 1987A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE STUDIESDEPARTMENT OF GEOGRAPHYWe accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIAJune 1994© Rachel A. Spronken—Smith, 1994In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.V (Signature)___________________________Department of_____________The University of British ColumbiaVancouver, CanadaDate lct9DE-6 (2/88)AbstractWhile there has been a long tradition for the integration of architecture and landscapeto improve the urban environment, little is known about the effect of urban parks onlocal climate. In this study the park effect is determined through an integrated researchapproach incorporating field measurements of the thermal regime and energetics of urbanparks, together with scale modelling of nocturnal cooling in urban parks.The research is limited to consideration of the park effect in two cities with differentsummer climates: Sacramento, California (hot summer Mediterranean) and Vancouver,British Columbia (cool summer Mediterranean). In both these cities, surveys of summertime air temperature patterns associated with urban parks confirm and extend previousfindings. In temperate Vancouver, the park effect is typically 1—2°C, rarely more than3°C, although it can be higher under ideal conditions. However, in a hot, dry city, theeffect is considerably enhanced with parks as much as 5—7°C cooler than their urbansurrounds.A comparison of the surface energy balance of small open, grassed parks in these twocities demonstrates the importance of evapotranspiration in park energetics. In hot, drySacramento, evaporation in the park was advectively—assisted and exceeded that at anirrigated rural site. Strong advective edge effects on evaporation were observed in thiswet park. These decayed approximately exponentially with distance into the park. Theurban park in Vancouver was moist, but unirrigated. While evaporation dominated thesurface energy balance, the sensible heat flux was positive through most of the day, andevaporation was not strongly influenced by advection. The evaporation trend in the parkprobably reflected the turbulence and soil moisture regimes. However, an irrigated lawn11in Vancouver did exhibit edge—type advection. This suggests the soil moisture regimemay be critical in determining whether evaporation exceeds the potential rate.The contribution of processes to nocturnal cooling in urban parks was determinedthrough scale modelling. It showed that surface geometry and the urban—park differencein thermal admittance may be of equal importance in nocturnal cooling. Parks withhigh sky view factors have increased radiative cooling and if the park is very dry (andtherefore has a low thermal admittance), the cooling is further enhanced. Evaporativecooling is critical in establishing the park as a “cool island” at sunset, but the presenceof moisture slows cooling through the night.Integration of the field and model data leads to the development of guidelines forplanners regarding the design of parks for maximum climatic benefit. The optimum sizeof the park depends to a large extent, on the geometry of the urban surrounds. Tomaximize radiative cooling, the width of open park areas should be at least 7.5 times theheight of the trees or buildings around the park border. Large parks increase the sizeof the volume of air cooled and this increases the potential for advection of cool air intothe neighbourhood. It is suggested that if cooling is the objective, the optimum designis a savannah—type park with loose clusters of trees interspersed by wide open, irrigatedgrass. The arrangement of trees must be chosen with great care to allow the advectionof air both into, and out of, the park.111Table of ContentsAbstract iiList of Tables ixList of Figures xiAcknowledgement xx1 Introduction 11.1 Definitions and scale considerations 21.2 Climatic amelioration by greenspace 61.2.1 Impact of greenspace at the microscale 61.2.2 Impact of greenspace at the local—scale 91.2.3 Impact of greenspace at the mesoscale 151.3 Rationale and research approach 181.4 Objectives 212 Survey of the Park Effect in Two Cities with Different Summer Climates 222.1 Method and Instrumentation 222.1.1 Survey areas 222.1.2 Data collection 242.1.3 Data analysis 272.2 Survey results 28iv2.2.1 Spatial and temporal patterns of park influence on air temperaturesin Sacramento 282.2.2 Spatial and temporal patterns of park influences on surface and airtemperature in Vancouver 332.3 Comparison of PCIs in cities from hot and cool summer Mediterraneanclimates 463 Conceptual Framework for Field Measurement 503.1 Spatial variability of SEB across a park 503.1.1 Influence of fetch on the SEB in a park 503.1.2 Influence of the turbulence regime on the SEB in a park 563.2 Site locations 643.3 An overview of techniques used to measure the SEB fluxes 664 Surface Energy Balance of an Urban Park in a Hot Summer Mediterranean Climate — the Sacramento Study 724.1 Observation programme 724.1.1 Physical setting and observation sites 724.1.2 Instrumentation 754.1.3 Probable errors in surface energy balance instrumentation . . . . 794.2 Urban park surface energy balance (SEB) 804.2.1 Energy balance partitioning 804.2.2 The influence of fetch on QE 874.2.3 Summary of the energetics of a suburban park in a hot summerMediterranean climate 884.3 The urban park in its context — park, suburban and rural comparisons 894.3.1 Energy partitioning in different land—use types 89V4.3.2 Summary of the urban park surface energy balance compared toother land—use types in a hot summer Mediterranean climate. . . 955 Surface Energy Balance of an Urban Park in a Cool Summer Mediterranean Climate — the Vancouver Study 965.1 Observation programme 965.1.1 Physical setting and observation sites 965.1.2 Instrumentation 995.2 Urban park surface energy balance (SEB) 1055.2.1 Background climate 1055.2.2 Energy balance partitioning 1065.2.3 The influence of fetch on QE 1115.2.4 Summary of the energetics of a suburban park in a cool summerMediterranean climate 1145.3 The urban park in its context — park, suburban and rural comparisons 1155.3.1 Energy partitioning in different land—use types 1155.3.2 Summary of the urban park SEB compared to other land—use typesin a cool summer Mediterranean climate 1215.4 Comparison of park energetics in different climates 1216 Scale Model Design for Nocturnal Cooling in Urban Parks 1266.1 Considerations in nocturnal park cooling 1266.2 Objectives of scale modelling 1286.3 Scaling considerations 1286.4 Scale models 1306.4.1 Simulation of radiative cooling. . 1326.4.2 Simulation of conductive cooling. 132vi6.4.3 Simulation of evaporative cooling 1346.4.4 Combined influences of geometry, thermal admittance and evaporation 1356.5 Measurements 1376.5.1 Surface temperature measurement by thermocouples 1376.5.2 Remotely sensed surface temperature measurements 1386.6 Analysis 1386.6.1 Corrections 1386.6.2 Theoretical framework 1406.6.3 Derivation of thermal admittance 1456.6.4 Calculation of surface temperature park cool islands, PCI8s . . . 1466.6.5 Replicability of results 1467 Scale Model Results of Nocturnal Cooling in Urban Parks 1497.1 Radiative cooling 1497.1.1 Influence of park size 1497.1.2 Influence of park type 1507.2 Thermal admittance effects on park cooling 1577.3 Evaporative effects on park cooling 1597.4 Combined effects on park cooling 1607.4.1 Radiative and thermal effects 1607.4.2 Relative contribution of processes to park cooling 1647.5 Comparison of scale model results with surface cooling observed in Vancouver parks 1688 Conclusions 1708.1 Summary of results 170vii8.2 Implications of research for park design 173References 176Appendix A Survey of the Park Effect in Tucson, Arizona 191Appendix B Errors in Scale Modelling 197B.1 Introduction 197B.2 Surface temperature errors 197B.3 Emissivity errors 199Appendix C Sky View Factor Calculations 201C.1 Calculation of ib for locations in grass parks and canyons 2010.2 Calculation of for locations in grass parks with tree borders 203C.3 Calculation of b3 in savannah, garden and forest parks 204vi”List of Tables1.1 Scale classification of the urban canopy layer as a framework for studyingthe climatic impacts of greenspace 52.1 The average PCIs for different park types in Sacramento in August, 1993.Four multi—use types, two irrigated grass, two dry savannah, one forestpark and one golf course were surveyed. All PCIs and UHI are in Celsius.Sunrise was about 520 LAT and sunset about 1835 LAT 293.1 Specifications of the mini—lysimeters 714.1 Daily and daytime (Q* > 0) averages and ratios for energy balance fluxesat sites in Orville Wright Park. All fluxes are in W m2 834.2 Daily and daytime 2—day averages and ratios of energy balance fluxes fordifferent land—use types in Sacramento, CA 905.1 Daily and daytime (Q* > 0) energy balance fluxes and and ratios for sitesin Trafalgar Park. All fluxes are 3—day averages expressed in W m2. . 1075.2 Daily and daytime (Q* > 0) energy balance fluxes and and ratios for sitesin Vancouver. All fluxes are four—day averages in W m2 1175.3 Comparison of daily energy balance fluxes (W m2), and ratios for theSacramento park (Orville—Wright), Vancouver park (Trafalgar) and anirrigated Vancouver lawn 1236.1 Thermal admittance of the materials used in scale modelling 134ix6.2 Estimated emissivities for model materials in both their dry and moiststates. The means and standard deviations (o) are derived from three tests. 1427.1 The effect of park type on the magnitude of the average PCI3. All PCIsare in Celsius 1577.2 The effect of surface thermal admittance differences (Lt) upon the PCI3.The “city” has no canyon geometry and has .t 366 J m2 s K—’.Units of all thermal admittances in J m2 s K—’ and all temperaturesin Celsius 1587.3 Absolute magnitude of the PCI3 for different model designs. All temperatures are in Celsius and values are rounded to the nearest integral value. 165A.1 Summary of the nocturnal thermal climate of parks in Tucson. City—widetraverses only occurred on May 15, May 24 and June 2. The remainingtraverses only surveyed Reid Park. All traverses began an hour after sunset. 193B.1 Summary of errors in surface temperature measurement by thermocouples. 198xList of Figures1.1 The energy balance of a soil—plant—air volume (from Oke, 1987) 31.2 The energy balance of the urban “surface” (from Oke, 1984) 31.3 The classification of urban “surface” boundary layers (Oke, 1984) 51.4 Scheme of daytime energy exchanges between an isolated tree and its streetcanyon environment (from Oke, 1989). T1 and Ta are leaf and air temperatures 71.5 Isotherms (°C) in Chapultepec Park, Mexico City, on December 3, 1970(about 600 h), in clear and calm conditions. The park area is shaded.(After Jauregui, 1991) 112.1 Location of the fixed sites in Sacramento, California, and the traverse routefor the park survey. The urban heat island traverse was from the dry ruralsite to downtown Sacramento 232.2 Location of the fixed sites in Vancouver, British Columbia and the traverseroute for the park survey. The urban heat island traverse was from therural site to downtown Vancouver 252.3 Air temperature traverse including several parks and open spaces in Sacramento at 2100 LAT on August 16, 1993. Sunset was at 1835 LAT 302.4 The PCI and UHI from fixed sites in Sacramento. The two—day average PCI, and UHI based on suburban—dry rural difference (SU—DR) andsuburban—wet rural difference (SU—WR) are shown in (a) while the nocturnal cooling after sunset is shown in (b) 32xi2.5 PCI range and mean for parks surveyed in Vancouver in summer, 1992.The data are averages from 3 survey periods with comparable weather.The average wind speed is also shown as a dashed line 342.6 Aerial photographs of Vancouver parks used in helicopter survey of surfacetemperatures 362.7 Surface temperatures of Vancouver parks and neighbourhoods at 1400 LATon August 24, 1992 372.8 Surface temperatures of Vancouver parks and neighbourhoods at 1930 LATon August 24, 1992. Sunset was at 1919 LAT 382.9 Surface temperatures of Vancouver parks and neighbourhoods at 2300 LATon August 24, 1992 392.10 Surface temperatures of Vancouver parks and neighbourhoods at 500 LATon August 25, 1992. Sunrise was at 518 LAT 402.11 Trace of air temperature from a traverse in Vancouver at 1900 LAT onAugust 24, 1992. Sunset was at 1919 LAT 412.12 Average PCIs for different park types in Vancouver on August 24/25, 1992.The survey comprised five grass parks (with varying soil moisture), twograss parks with large tree borders, a garden and a multi—use park andone savannah park 422.13 Isotherms (°C) for air temperature around Trafalgar and Prince of WalesParks and neighbourhoods, August 24 and 25, 1992. The data were gathered by bicycle surveys within and around the parks 44xl’2.14 PCI and UHI, and nocturnal cooling for fixed sites in Vancouver. The dataare four—day averages (July 25—28, 1992). The PCI is a suburban—park airtemperature difference and the UHI is a suburban—rural air temperaturedifference (a). Nocturnal cooling for different land—use types is shown in(b) 473.1 Postulated variation of the latent flux with distance of fetch for an irrigated(a) and (b) an unirrigated park 543.2 Flow distortion over a building — the cavity and tubulent wake zones.(From Hosker, 1984 based on Woo et al. 1977 and Hunt et al. 1978.) . 573.3 Flow distortion over a barrier — the quiet and turbulent wake zones. (AfterMcNaughton, 1988.) 593.4 Eddy diffusivity patterns in a forest clearcut (from Chen et al., 1994). . 613.5 Scenario for likely impacts of both fetch and the turbulence regime on thespatial variability of the daytime SEB in a suburban park. The length ofthe arrows indicates the approximate proportions 643.6 Design of the miniature lysimeter (after Crimmond and Isard, 1992). . 714.1 Photographs of each study site in Sacramento 734.2 Sampling plan for surface energy balance in Orville Wright Park 764.3 Differences in diurnal air and surface temperature, Ta and T8 (°C), relativehumidity, RH (%) in (a), and vapour pressure, LSVP, in (b), betweenadjacent park and paved sites at Orville Wright Park, Sacramento 81xlii4.4 Two—day average surface energy balances at three park sites in OrvilleWright Park, Sacramento in August, 1991. The net radiation and sensibleheat fluxes measured at Site 3 in August 1993 are also shown for comparison. For the 1991 measurements, net radiation was within 10% of theaverage on an hourly basis; QE within 20% and QH and QG within 30%.In 1993, Q* varied less than 4% and QH was within 25% of the average onan hourly basis 824.5 Comparison of the measured QE with Q and QEp at three sites in OrvilleWright Park, Sacramento in August, 1991 854.6 Two-day average of the SEB flux ratios for daytime hours at three parksites. To facilitate interpretation, the data for Site 1 near midday areomitted due to shading by nearby trees 864.7 The variation of average daytime QE with distance of fetch from the parkedge for August 27 (YD 239). The average wind speed for this period was2.84 m s at the nearby suburban site 884.8 Average energy balances for different land—use types in Sacramento, CA.,based on two days observations in August, 1991. Net radiation is within10% of the mean on an hourly basis; QE within 20%; and QH and QGwithin 30% 914.9 Cumulative evaporation for the different land—use types in Sacramento,CA. Data are two—day averages 924.10 Daytime flux ratios for different land—use types in Sacramento, CA. Dataare two-day averages. The Bowen ratio for the dry rural site is not shownas it is about 60 945.1 Photographs of study sites in Vancouver 97xiv5.2 Location of sites in Trafalgar Park, Vancouver 1005.3 A comparison of QE as measured by a mini—lysimeter and estimated asthe SEB residual for Site 5 in Trafalgar Park 1025.4 Time trend of daytime average QEr/QEres for Sites 1-5 in Trafalgar Park,Vancouver 1035.5 Surface forcing for evaporation in Trafalagar Park, Vancouver. The surfacetemperature difference between the park and the adjacent gravel field isshown in (a), with the vapour pressure difference between the near—surfacepark air and the overlying air shown in (b) 1065.6 Surface energy balance for Sites 1-3 and 5 in Trafalgar Park, Vancouver.Data are three—day averages. There is considerable variability of the fluxesabout their mean because the weather was clearing. On an hourly basisthe fluxes may vary by up to 30—40% of their mean value 1085.7 Average daily cumulative evapotranspiration for Sites 1—3 and 5 in Trafalgar Park, Vancouver 1095.8 Surface energy balance flux ratios at sites in Trafalgar Park. Data arethree—day averages 1105.9 The variation of QE with distance of fetch from the park edge based onthe daytime average for July 24-25, 1992. The average wind speed for thisperiod was 2.19 m s1 1115.10 Possible scenario for the turbulence regime in Trafalgar Park, Vancouver 1125.11 Average soil moisture for the observation period in Trafalgar Park, Vancouver 114xv5.12 Surface energy balances for the suburban, park and rural sites in Vancouver. Data are four—day averages. On an hourly basis net radiation iswithin 10% of its average, while the turbulent and storage fluxes may varyup to 20—25% 1165.13 Daytime fluxes normalized by net radiation for the suburban, park andrural sites in Vancouver. Data are four—day averages 1195.14 Comparison of surface forcing on evaporation between the Sacramentoand Vancouver parks and their suburban environs. The surface temperature difference between urban and park surfaces are in (a), while (b) isthe vapour pressure difference between the near—surface park air and theoverlying air 1225.15 Comparison of park SEB fluxes normalized by their maxima (i.e. Q/Qmax).Data are ensemble averages from the Sacramento and Vancouver studies. The normalizing maxima values (in W m2) for Orville Wright Park,Sacramento are: = 529, QE = 435, QH = 110, QG = 24; and forTrafalgar Park, Vancouver: Q* = 480, QE = 311, Qg = 115, QG = 56. . . 1246.1 Scale model design to simulate radiative cooling in different park types.The “grass park” is open with no “trees”. The park dimensions are0.36 xO.36 m and the buildings are 40 mm high and wide 1336.2 Scale model design to simulate conductive cooling in urban parks 1356.3 Scale model design to simulate evaporative cooling in urban parks 1366.4 Comparison of surface temperature estimates from the AGEMA and thermocouples. Data are averages for three model runs 139xvi6.5 Sensitivity tests of the difference between assumed and actual emissivityon derived temperatures from AGEMA images: (a) shows the temperaturedifference when = 0.97 is assumed; (b) shows the temperature differencewhen = 1.0 is assumed 1416.6 Comparison of surface cooling observed in the scale model (flat fir surface)with that predicted by the SHIMpDE model when initialized by scale modeldata. The observed data are the averages of five model runs. Standarderrors for the observations are indicated 1456.7 Test of the replicability of the scale model to simulate cooling in a “grassedpark”. Average urban and park cooling curves (a) are shown together withthe derived PCI8 (b). The averages and their standard errors derived fromfour replicates 1487.1 Map of sky view factor for a quadrant of the large open grass park model.The surrounding buildings are 40 mm high. The location of the thermocouples used in the analysis is also shown 1517.2 The influence of park size on development of the PCI3 after sunset. . . 1527.3 Relation between the ratio park width : building height and sky viewfactor in the centre of the park. Building height was held constant, butcan in practise, vary 1527.4 Influence of park type on cooling of different facets: (a) open grassed parks;(b) grass parks with tree borders; (c) savannah parks; (d) garden parksand (e) forest parks. Cooling rates in the urban canyon are shown forcomparison. The ,b3 for each facet is given in brackets 153xvii7.5 Park surface temperatures at the end of the model “night” (625 s). (a)open grassed parks; (b) grass parks with tree borders; (c) savannah parks;(d) garden parks and (e) forest parks 1557.6 The influence of park type on: (a) the growth of the PCI. (average b3 isshown in brackets) and (b) the relation between the average b3 and NPCI3.1567.7 PCI3s for open parks with differing thermal admittances 1587.8 Surface cooling (a) and POTs for parks with evaporating surfaces (b). Atsunset both parks had PCI3s of about 24°C. Cooling of the water due toevaporation only, is shown by E, while cooling due to thermal admittanceeffects is shown by p 1617.9 Surface temperatures after 625 s of cooling for parks under the canyons,dry park design: (a) LDF; (b) OSB; (c) plywood; and (d) concrete. . . 1627.10 The maximum intensity of the PCI3 in relation to the sky view factor, b3(shown in brackets) of the park surface, and the difference between thethermal admittance of the city and park environments (Lp) 1637.11 A comparison of POTs for different materials: (a) LDF; (b) OSB; (c) plywood; under the different model designs (flat, dry; flat, wet; canyons, dry;canyons, wet). In the “canyons” model designs, the open treeless park issurrounded by canyon geometry 167A.1 Location of the fixed sites and the traverse routes for the park surveys inTucson 192A.2 Air temperatures for the downtown to desert traverse an hour after sunset,May 24, 1991 195A.3 Isotherm map for Reid Park, May 22. Arrows indicate the direction oftraverse and the spots are check—points for the traverse 196xviiiC.1 Definitions of elements and coordinate system for calculating the viewfactor of a wall for a given point P on the ground 202C.2 Definitions of elements and coordinate system for calculating the viewfactor of a deciduous tree for the point P 203xixAcknowledgementFirst and foremost, many thanks to my supervisor Tim Oke. He has continually stimulated and guided me throughout my research and has always been a source of greatinspiration. I would like to express my appreciation of my supervisory committee: AndyBlack, Mike Church and Douw Steyn for their useful comments and prompt reading ofmy thesis.Special thanks to Jamie Voogt for patience under my never—ending enquiries, andto Arthur McLean. Thanks also to Dr. Sue Grimmond (Indiana University) for usefulinput throughout the field component. Several others assisted with field work: StephanieSmith, Trevor Newton, Katrina Richards and Brian Rehwald. Thanks also to CatherineGriffiths and Jessica North who helped put the finishing touches on the thesis.The Vancouver Park Board provided access to Trafalgar Park and the Mission OaksPark and Recreation Board in Sacramento were incredibly helpful during all stages offield work in Orville Wright Park. B.C. Hydro provided the site for the suburban towerin the Mainwaring substation.Funding for this research has been provided to Dr. T.R. Oke by Natural SciencesEngineering Research Council of Canada and Atmospheric Environment Service of Environment Canada. Personal funding was provided through the New Zealand Meteorological Service, and the University of British Columbia Graduate Fellowships and Teachingand Research Assistantships in the Department of Geography.Finally I would like to thank Harold. He has endured years of my problems; comeout at all hours of the night, and been very patient to the end. His support has beenterrific and has kept me going through some trying patches.xxChapter 1IntroductionThere is a long tradition in human history involving the integration of architecture andlandscape to improve the urban environment. It is not clear whether the early use ofvegetation in urban areas represented a conscious attempt to enhance the environmentor whether the improvements realized were merely accidental consequences of landscapemanipulations performed for other reasons (Hutchinson et al., 1983). Urban vegetationimproves the urban environment in a variety of ways through both social (e.g. aesthetic,psychologic, economic) and physical (e.g. climatic, air quality, hydrologic, biogeographic,acoustic) means. This thesis concentrates on the climatic benefits of urban vegetation butin so doing, the intention is not to downplay these other benefits. Indeed the attainmentof maximum climatic benefit may be at the expense of other benefits and in the finalanalysis, the full costs and benefits of greenspace must be weighed.The climatic benefits of urban greenspace (which includes private gardens, streettrees, parks and undeveloped or derelict vegetated land) have long since been realizedat least in a qualitative sense. Despite this recognition there has been little researcheffort as urban greenspace has not been a target for basic research. Consequently therehas been little coordinated understanding based on field observations of meteorologicalprocesses or effects (Oke, 1989).11.1 Definitions and scale considerationsit is important to maintain a sense of scale and anappropriate definition of the surfacewhen studying atmospheric interaction between greenspace and the urban surrounds. Ina sense, greenspace may be thought of as a rural surface in urban surrounds. In a ruralsetting the meteorological definition of a surface isgenerally straightforward, provided itis homogeneous and extensive. If there is a substantial plant canopy, energy exchanges involve a soil—plant—air volume (Fig. 1.1). Thus the near—surface energy balance (hereafterreferred to as SEB) may be written:QH + QE + zQs + LiQp + LQA (1.1)where the terms are the flux densities and is afinite difference approximation (i.e.the difference or net change in a quantity):Q* net all—wave radiation; QH the turbulentsensible heat flux; QE the turbulent latent heat flux; Qs the net sensible heat storageby components in the system; ZQp the net biochemical heat storage due to plant photosynthesis; ZIQA is an advection term for net energy gain or loss due to sensible heat andlatent heat transport. The net biochemical heat storage is usually negligible for cropsand grass surfaces and for such canopies the ground heat flux, QG is used to approximateIQs, assuming the canopy does not have a high storage capacity.The SEB of greenspace, by virtue of being locatedin an urban environment, willinteract with the SEB of the urban surface. Defining the urban surface is a more complexproblem given the level of heterogeneity. Oke (1988) suggests theuse of a near—surfaceactive layer or volume, analagous to the soil—plant—air volume above (Fig. 1.2). Theurban volume extends from below the ground surface to roof—top level. Anthropogenicheat input (QF) is added to the energy balance so that the SEB is given:Q* + QF QH + QE + Qs + LQA. (1.2)2Q* QE QHLoweratmosphereTCanopy +stem layerSoil*Figure 1.1: The energy balance of a soil—plant—air volume (from Oke, 1987).-Figure 1.2: The energy balance of the urban “surface” (from Oke, 1984).3The energy release due to human activities depends on the city size and on the per capitaenergy use which is influenced by climate, the degree and type of industrial activity,the site of electricity generation, the urban transportation system, etc. (Oke, 1984). Indensely populated cities in cold climates QF can be substantial and even surpass theradiative energy input in winter. In summertime studies this flux is sufficiently small tobe regarded as negligible.This definition of the urban surface conveniently coincides with an urban classificationdeveloped by Oke (1976, 1984) (Fig. 1.3). The layer below roof—level is called the urbancanopy layer (UCL) and is dominated by the microscale effects of the site characteristics.Above roof—level is the urban boundary layer (UBL) which includes the roughness layerimmediately affected by roughness elements, the turbulent surface layer and the mixedlayer.A framework for studying climatic impacts of greenspace is developed from Oke (1984,1989) (Table 1.1). At the microscale, greenspace can impact the climate of buildings andcanyons; at the local—scale it impacts the climate of city blocks, while at the mesoscale theurban forest (which includes all woody vegetation) influences the climate of the land—usezone or the city.In assessing the climatic impact of greenspace, careful consideration must be givento macroscale climate, as well as the scale of influence within the urban canopy. Themacroscale climate is the climate background within which the city is located. Dependingon the macroscale climate, greenspace can be manipulated in various ways to improveurban climate.4(a)I— Mixed layerPBLUBL——/ Surface layerY-.R8L.1:%imSurface layer-——-. Roughness layer— —CL________________________________________________—Figure 1.3: The classification of urban “surface” boundary layers (Oke, 1984).Table Li: Scale classification of the urban canopy layer as a framework for studying theclimatic impacts of greenspace.Unit Built feature Creenspace Climate Dimensions Scalefeatures phenomena H W Lbuilding building tree, garden wake, plume 10 10 10 m microcanyon street boulevard thermal climate 10 30 300 m microshelterbelt shadeblock city block, park, wood local breezes,— 0.5 5 km localfactory park coolingland—use residential, greenbelt air quality and— 5 5 km mesozone industrial suburban forest climate districtscity built—up area urban forest heat island,— 25 25 km mesocity breezes51.2 Climatic amelioration by greenspaceUrban greenspace can have a marked effect on urban climate. Early attention focused onthe role of vegetation in removing pollutants from city streets. Parks were “the lungs” ofcities (Brezina and Schmidt, 19371). in recent years, research has started to assess therole of greenspace in modifying climate parameters such as air and surface temperatures,humidity and wind. Studies typically focus on the microclimatic effects of greenspacewith little consideration to its effects at the mesoscale. Comprehensive reviews of progressin the field are given by Hutchinson et al., (1983) and Oke, (1989).1.2.1 Impact of greenspace at the microscaleThe major influence of vegetation on microclimates occurs through the interception ofradiation (i.e. shading) and the deflection of winds by plant canopies (Hutchinson et al.1983). The daytime energy balance of a tree (Figure 1.4) is discussed by Oke (1989).Tree leaves intercept, reflect, absorb and transmit solar radiation. Their effectivenessdepends on the density of foliage, leaf shape and branching patterns. The largest effectcan be gained from trees with tall, dense and wide canopies. Trees surrounded by builtenvironments receive large amounts of reflected short—wave radiation. The long—waveradiative energy input is also greatly enhanced because of adjacent warm surfaces. It isalso possible for the air temperature in a street canyon to exceed the leaf temperature,subjecting it to a sensible heat load by advection. Heat is dissipated primarily throughevapotranspiration. Oke (1989) points out that this process is fundamentally dependenton the water balance and wind climate of the tree. The water supply may be restrictedand/or contaminated. The stomata may be physically blocked by particulates or theexcessive heat load may lead to stomatal closure. Furthermore, wind may be sheltered1Cited in Landsberg, (1981).6Figure 1.4: Scheme of daytime energy exchanges between an isolated tree and its streetcanyon environment (from Oke, 1989). T1 and Ta are leaf and air temperatures.in street canyons which reduces ventilation.Vegetation reduces air temperature by day mainly through the provision of shadeand evaporative cooling. Deciduous trees are instrumental in heat control in temperateregions. During summer they intercept solar radiation and lower ground temperatures,while in winter, when they are leafless, they provide pleasant warming effects by enablingmore insolation to reach the ground. This effect though, has often been overestimated.Heisler (1982, 1986) found insolation reduced by as much as 80% by medium—sized deciduous trees in leaf, while in winter, insolation was reduced by 40—50%. Deciduous treesin leaf usually have a higher aibedo (about 20%) compared to coniferous trees (5—15%)(Oke, 1987). Therefore deciduous trees are more effective in reflecting incoming radiation.However this effect decreases in winter.Vegetation can also reduce temperatures by evapotranspirational cooling. Perhaps“cooling” is the incorrect term; rather it is a “prevention of heating” as available energyheat gainsdirect and diffuselong-wave short-waveheat losseslatent heat7is channelled into the evaporation of water (latent heat) rather than sensible heat. Lowry(1988) estimated evapotranspiration cooling rates from six trees along 10 m of a streetcanyon with a lOx 10 m cross section. He assumed an average leaf transpiration rateof 70 W m2 at midday and a canyon ventilation rate of 100 volumes per hour. Thisresulted in a cooling rate of 0.3°C h1. However, Oke (1989) comments that even this“passive cooling” of isolated trees may not be significant in reducing air temperatures atground level because most of the transpiration occurs at the top of the trees and maynot be mixed throughout the volume.The combined effects of shade and passive cooling can significantly reduce surfaceand air temperatures. In a descriptive survey of summertime urban surface temperatures, Kondo and Suzuki (1983) found that the presence of roof—gardens reduced surfacetemperatures of houses by up to 40°C. Measurements from Sacramento, California, suburbs with mature canopies found daytime air temperatures 1.7—3.3° C lower than in openareas (Taha et al., 1988). In Miami, Florida, average summer daytime air temperatureswere reduced by 3.6°C under a large tree (Parker, 1989). In higher latitudes the effectof trees on lowering air temperatures is less. Souch and Souch (1993) report decreases of0.7—1.3°C under a variety of trees in Bloomington, Illinois.Given the potential of greenspace to decrease surface and air temperatures, severalstudies have tried to assess the energy savings of different landscape designs. McPherson(1989) used scale model houses to assess cooling and water impacts of different landscapes.He found that the high costs of maintaining a lawn outweighed the benefits from lowertemperatures due to evapotranspirational cooling. However, the provision of shade andpassive cooling by shrubs reduced energy used in air—conditioning by 30%. SimilarlyParker (1981, 1983, 1987) used vegetative landscaping to reduce energy use in mobilehomes. Substantial air—conditioning savings were made — up to 76%.Such research has spawned efforts to model the effects of landscaping on residential8energy use. Thus microclimate models are linked with building energy simulations toquantify the effects of vegetation on energy savings. In simulations for Sacramento,California and Phoenix, Arizona, Huang et al. (1987) estimate that a 25% increase inurban tree cover can save 40% of annual cooling energy use in residences. Comparablesavings in residential energy consumption were found for summertime simulations in largeCanadian cities (Akbari and Taha, 1992).The microscale shelter provided by vegetation affects pedestrian comfort and safety,air pollution dispersion and energy conservation (Oke, 1989). Trees and shrubs controlthe wind by obstruction, deflection and filtration. The effect and degree of control depends on species size and shape, foliage density and retention and the actual placementof plants (Grey and Deneke, 1978). Dense plantings can provide relief from cold winterwinds, but in warm climates they can impede air flow and decrease human comfort by reducing both convective and evaporative heat loss (Heisler, 1977). In summary, therefore,each application requires careful consideration.1.2.2 Impact of greenspace at the local—scaleThere is a paucity of research on the local—scale effects of greenspace beyond descriptivesurveys. It is well documented that urban parks can be cooler than their surroundingenvironment. Often the influence of parks on air temperatures is noted during urban heatisland surveys. Few surveys assess the park effect itself. This effect can vary dependingon the park type, the nature of the urban surrounds and the macroscale climate. Mostsurveys are in mid—latitude cities. Few studies are reported from low—latitude cities witheither hot, dry or hot, humid, climates.Parks can be classified by considering the arrangement of vegetation:• Grass — open grass surface;9• Crass with tree border — open grass surface with a tree border;• Savannah — grass surface with trees interspersed throughout;• Garden — cultivated park with a mixture of grass, trees, shrubs and flowerbeds;• Forest — park that has a continuous tree canopy coverage;• Multi—use — park that has many different components e.g. grass playing fields,treed areas, swimming pools etc.• Golf courseIn addition to this simple classification the soil moisture regime of the park should benoted. Surveys of the thermal effect of parks should also comment on the nature of theurban surrounds. If the park is surrounded by a commercial area that is densely builtwith few pervious surfaces, its effect is likely to be enhanced. However, if it is surroundedby a well—treed residential area, the influence of the park on local temperatures will bereduced.Only a few air temperature surveys are reported from hot, dry cities. This is surprisinggiven the potential of parks to cool neighbourhoods. Kirby and Sellers (1987) studied airtemperatures in Tucson, Arizona, during the period from late fall to early spring. Theyfound that cold air drainage plays a dominant role in determining the fall and winterthermal climate of the city region, but they also note low temperatures centred on thelargest park in the city.Jauregui (1973, 1991) presents results from one of the few studies of thermal patternsassociated with urban parks in a low latitude city with a tropical highland climate. Ina study of a large park (525 ha) in Mexico City, he found maximum cooling during thedry season at which time minimum temperatures were 3—4°C cooler, and the extent ofcooling reached about one park width into the urban neighbourhood (Fig. 1.5).10kilometresFigure 1.5: Isotherms (°C) in Chapultepec Park, Mexico City, on December 3, 1970(about 600 h), in clear and calm conditions. The park area is shaded. (After Jauregui,1991).11Studies of thermal regimes in hot, humid cities have indicated the park effect maybe substantial. In Kuala Lumpur parks may be 4—5°C cooler than nearby commercialareas in the afternoon (Sham, 1986, 1987, 1991). Sham suggests that relatively smallclumps of shady trees in the commercial area of a city can be far more effective than largegrass—covered fields with isolated trees in moderating afternoon temperatures. Howeverin the evening the well—treed parks are only slightly cooler (about 1°C) than their urbansurrounds. In tropical Madras, India, Sundersingh (1991) reports parks may have anocturnal cooling influence of up to 4°C.In these hot, humid cities well—treed parks have an important role in lowering daytimeair temperatures probably mainly through the provision of shade. In this humid environment the evaporative cooling of greenspace may be of reduced importance to that of a.dry city. There is also a detrimental impact of transpiring vegetation on human comfortthrough increased humidification. In a moist environment, there will be several sourcesof moisture for evaporation other than urban parks (e.g. undeveloped lots, interceptionof precipitation by buildings, pervious surfaces etc.). Thus it is probable that sensibleheat is less in hot, wet cities, with a higher latent heat flux than is observed in dry citieswhere moisture sources are much more limited. Therefore it is the macroscale climate,rather than any differences in greenspace characteristics, that determine the potentialpark effect. However, the park type, within the macroscale climate is of importance inrealizing this potential.Several park surveys have been conducted in mid—latitude cities: New York andSyracuse, N.Y. (Herrington et al., 1972; Herrington, 1977); London (Chandler, 1965);Montréal, P.Q. and Vancouver, B.C. (Oke, 1989). The thermal contrast between thepark and urban surrounding is best developed on calm and cloudless nights (as for theurban heat island). The nocturnal coolness of the park is established soon after sunsetfollowing which the park and urban area cool at similar rates (Oke, 1989). For parks12ranging in size from 29 to 500 ha, results show that nocturnal temperatures inside thepark are rarely more than 3°C, and typically 1—2°C, lower than the surrounding urbanarea. Larger thermal influences (up to 6°C) are reported from parks in Göteborg, Sweden(Eliasson, 1993; Lindqvist, 1992). The park cooling establishes a zone of larger influencebeyond its borders. This zone of lower temperatures associated with the park is hereafterreferred to as a “park cool island”2In warm, dry summer climates it is likely that the park effect will be greater thanin a warm, moist climate. The influence of parks is probably similar (but reduced inmagnitude) to that in hot dry cities. This is due to the relative warmth of the city ratherthan any inherent difference in the greenspace. In warmer, dry cities, any moisturesources such as irrigated greenspace are likely to have a greater effect on the thermalregime of the urban environment.As well as producing “cool islands”, urban parks are characteristically “moistureislands”. The humidity difference between urban and natural surfaces tends to be smalland the spatial pattern is often complex. Given the similarity between park and ruralsurfaces it is useful to consider urban—rural humidity differences as an indicator of likelyeffects in a park. The consensus of mid—latitude studies is that urban canopy air isusually drier by day than rural air but slightly moister at night. This pattern is mostpronounced during fine summer weather (Oke, 1987). Higher humidities by day overnatural areas can be attributed to transpiring vegetation. Over open grassed areas thiseffect is small, but under a forest canopy, humidity is usually higher. While severalresearchers have noted the increased humidity in parks compared to urban surrounds,few studies give quantitative estimates. In the early evening parks cool more rapidly than2This term is not strictly correct in Oke’s framework of heat island terminology. He likens the typicaltemperature trace from the rural surrounds to the city centre to a warm (urban) island set amongst acold (rural) sea. Parks thus form depressions (or cool pools) in the warm urban landscape. However, theterm “cool island” is in widespread use and is appropriate for parks given the biogeographical analogywhere “is’ands” are remnants of vegetation.13their urban surrounds. This may lead to convergence of moisture in the lower layers of theatmosphere as the evapotranspiration from the surface exceeds the loss to higher layersdue to dampened turbulence. Thereafter, humidity decreases through the night and avapour inversion may form and dewfall occurs. On the other hand, over built surfaces,the lack of evaporation, reduced dewfall, anthropogenic vapour and the stagnation ofairflow all combine to maintain a less humid atmosphere.Vegetation in parks through its effect on the wind regime, can lead to either a decreaseor an increase in local air temperatures. Wind can increase evaporative cooling and theprocess of advection can carry cool air beyond the park boundaries. However, vegetativewindbreaks can lead to a decrease of velocity and the creation of a sheltered zone withinwhich air temperatures are increased. Therefore it is important to plan the configurationof the vegetation to achieve the desired effect.There have been several cursory observations and suggestions of local park breezes(eg. Whiten, 1956; Gold, 1956; Bernatzky, 1982). In clear, calm conditions Oke (1989)suggests that parks develop their own climate in situ. A cool park may thermally inducea pressure gradient leading to a divergent outflow of cool air at low levels (park breezes)and subsidence over the park. This flow may explain the outward diffusion of cool air fromits source that has been observed in park surveys (e.g Jauregui, 1991; Oke, 1989). In lightwinds, advection of cool park air can similarly result in the lowering of air temperaturesin neighbourhoods downwind.There have been a few attempts to numerically model the influence of greenspaceat the local—scale. O’Rourke and Terjung (1981) combined a multi—layered canopy leafenergy budget model and a complex street canyon energy budget model. They simulated energy balances for four urban designs that incorporated parks and roof gardensin different building morphologies. The addition of greenspace increased net short—waveradiation and net radiation and decreased system re—radiation and sensible heat flux. A14numerical turbulence model used by Honjo and Takakura (1991) estimated the coolingeffects of greenspace on urban surroundings. In the simulation, the scale and locationof the park was changed. Advection of cool air from a park 100 m across resulted inlower air temperatures up to 300 m downwind. As the park size increased to 400 m theinfluence only increased to 400 m downwind. Their results suggest that smaller greenareas (100 m across) with sufficient intervals (about 400 m) are preferable for effectivecooling of surrounding areas.1.2.3 Impact of greenspace at the mesoscaleThe effects of greenspace on the urban meso—climate are little understood (Herrington,1976). Givoni (1991) suggests that there are limited effects of greenspace beyond themicroscale. Nevertheless much research is directed at assessing the use of the urbanforest to mitigate urban heat islands (e.g. Akbari et al., 1988; Dwyer et al., 1992). Littleattention has been given to other climatic impacts of the urban forest except in Oke’sreview (1989).The urban landscape has aerodynamic roughness similar to forests (0.5—5 m) (Oke,1987). There may be a seasonal variation in roughness; with greater values when deciduous trees are in leaf. This seasonal leaf change can also alter the roughness environmentby removing some existing scales of regularity (streets are lined by trees with heightssimilar to houses), establishing new regularities (tree lines taller than the houses) orcreating irregularities (scattered tall trees and clumps in an otherwise uniform array ofbuildings) (Oke, 1989).There is remarkably little variation in radiation exchange across urban areas despitethe wide range of surfaces found at the microscale (Oke, 1989). Greenspace that contains open grassland generally has higher albedos than either suburban or urban surfaces.15Brest (1987) estimated albedos for different land—uses in Hartford, CT. Downtown areas had albedos from 8—12%; suburban — 11—17%; treed greenspace — 7—19%; non—treedgreenspace — 17—23%. He noted that vegetated surfaces have a higher albedo than mosturban surfaces and exhibit seasonal patterns in reflectance associated with their phenology. However Vukovich (1983) found no difference in reflectivity between Forest Park (asavannah park) and adjacent neighbourhoods in St. Louis.While greenspace may have a higher albedo, the emission of long—wave radiation isreduced due to passive cooling. Therefore, although net short—wave radiation may bereduced, net long—wave radiation is increased thus offsetting changes in net radiation.Hence there is little variation in net radiation across land—uses with varying amountsof greenspace. The conservative nature of net radiation has been noted by Schmid eta!. (1991) who found less than 5% difference over suburban surfaces whose greenspacecoverage varied substantially (35%—85%). At the mesoscale, White et al. (1978) foundless than 10% difference in net radiation from different land—uses in St. Louis.Oke (1989) suggests that it is the “hydrometeorological role of the urban forest thatcommands most attention”. A major problem of urbanization is the waterproofing ofsurfaces which results in rapid runoff. Urban greenspace plays an important role inintercepting and storing precipitation, thus slowing runoff. The other major hydrometeorological role of urban vegetation is transpiration which has important impacts on themoisture and thermal regimes of urban areas.Recent measurements of the latent heat flux in suburban areas has shown that evapotranspiration can be substantial (30—50% of net radiation) (eg. Roth, 1991; Grimmondand Oke, 1994). This considerable water vapour flux is probably derived from irrigatedgardens and parks, and trees tapping deeper soil water. Micro—advection of sensible heatfrom impervious surfaces may enhance the water loss by providing an additional energysource to net radiation. Advectively—enhanced evaporation was reported by Oke (1979)16in surface energy balance measurements over a suburban lawn in Vancouver, B.C.While a single transpiring tree may not affect a significant reduction in air temperatureat ground level, the collective effect of transpiring trees can affect air temperature at themesoscale. Oke (1989) suggests the rate of heating declines with greenspace and thatevaporative suppression (i.e. the thermal equivalent of energy used in evaporation whichwould otherwise contribute to turbulent warming) is of the same order of magnitude asthe heating. Thus greenspace can generate significant intra land—use thermal differences.Many numerical models have been used to assess the impact of greenspace on urbanclimates. Both Myrup (1969) and Rauner and Cernavskaja (1972) developed energybalance models to assess the influence of greenspace on the thermal regime of the city.They related temperature reductions to the amount of greenspace and both found acritical amount (20—30% cover) below which there was no amelioration of mesoclimate byvegetation. McElroy (197O), used a numerical model to simulate the nocturnal thermalstructure over a city and good agreement was found with observations in Columbus,Ohio. The introduction of greenbelts resulted in marked cooling but it was localizedboth horizontally and vertically.The model of Huang et al. (1987) (see section 1.2.1) has been used to project savings inenergy use from greenspace at the mesoscale. Indeed Akbari et al. (1988) use projectionsfrom this model to indicate potential savings at the national level. They suggest thatif every other single—family dwelling has three trees, annual energy use for the UnitedStates could be reduced by 30 billion kWh — an annual saving of about 2 billion dollars!Predictions from these models should be used with caution as their treatment ofevapotranspiration is very crude (Oke, 1989). Ross and Oke (1988) tested three modelsand found that none of them could simulate evapotranspiration satisfactorily.3Cited in Oke, (1989).17Greenspace alters the climate of the urban environment. The microscale effects ofgreenspace have been well studied and careful landscape design can achieve significantreduction of temperatures and alter the wind regime. The local—scale effects of greenspaceare only known in a qualitative sense, with few process—based studies. This highlightsa need for basic research on the surface energy balance and climate of urban parks.There is also a lack of research on the influence of greenspace at the mesoscale. Therehave been many attempts to model the effect of greenspace but they fail to simulateevapotranspiration realistically.1.3 Rationale and research approachAt the local—scale, effects of greenspace are poorly understood. Oke (1989) notes thatconsidering the potential importance of parks in modifying urban climate, it is surprisinghow little research has been done beyond descriptive surveys of effects. The coolingpotential of parks warrants further research. For cities with hot or warm summers,urban parks can be used to cool neighbourhoods, particularly at night when temperaturereductions of only a few degrees can be critically important to high—risk groups suchas the elderly (Clarke and Bach, 1971; Höppe 1991). The effect of parks on the urbantemperature field has not been well described. Most studies are in temperate mid—latitudecities; few are reported for cities in hot—dry or hot—humid climates. Little is known aboutthe diurnal influence of parks on the thermal regime. Further documentation of parkeffects on urban thermal climates is necessary, particularly for cities in hot climates.These issues raised by Oke (1989) and others (e.g. Herrington, 1976; Hutchinson eta!. 1983) highlight the need for process studies of urban parks. It is not known howurban parks modify the surface mass and energy balances or how they impact theirurban surroundings. Yet an understanding of the SEB of urban parks is fundamental to18an understanding of park climate. This calls for field research with energy balance andclimate measurements in urban parks. To assess the park influence fully, park climateand energetics should be compared to that of the surrounding land—use. To determinethe degree to which parks behave as natural or rural surfaces, rather than urban (built)surfaces, comparisons should be made between the energetics and climate, of park, urbanand rural environs.Although the cooling effects of parks are well known, the primary processes for parkcooling have not been elucidated. Oke (1989) raised the questions: Why is the within—park cooling not greater and why is it not similar to the rural case? Noting the nocturnaltiming of the maximum temperature differential between the park and city, Oke suggestedthat the relative failure of the urban system to cool may be just as important as anyspecial park cooling mechanism. However he felt it wise to await process studies beforeassigning energetic causes.Scale modelling is an attractive tool to determine the relative contribution of coolingprocesses. It can overcome the inevitable complications and limitations of field studiesand provides a valuable means to obtain experimental control. In addition, it is possibleto manipulate park characteristics such as size and vegetation type and assess theirimpact on the magnitude and extent of cooling.Scale modelling has been used successfully on a few occasions to simulate heating andcooling of urban surfaces. Oke (1981) used a simple hardware scale model to simulatenocturnal cooling rates for rural and urban surfaces. Surface cooling was simulated forcalm and cloudless nights as the urban heat island is best developed in such conditions.Surface geometry was altered to simulate radiative cooling of a rural surface (flat) andseveral different urban surface (canyon) geometries. The influence of thermal admittanceon cooling rates was also investigated by comparing cooling for a flat concrete “city”versus a wooden “countryside”. The model reproduced many features of the temporal19development of the urban heat island and compared favourably to field observations.Johnson and Watson (1987) used a simple scale model to simulate nocturnal coolingin urban canyons in order to validate a numerical model. They created a simple (upsidedown) urban canyon by placing a plywood sheet, representing the floor of a canyon, onthe open top of a freezer chest, whose walls act as canyon walls and whose floor is thecold, night sky. The modelled temperatures were in good agreement with the observedtemperatures.To study the park influence on local climate, the research strategy must consider bothdaytime and nighttime conditions. By day, radiant and turbulent processes are importantin controlling park climate. This requires measurements of surface energy balance andclimate parameters in parks, with comparisons to surrounding (urban), and rural sites.By night, turbulence is suppressed and radiative and conductive processes are probablythe important controls on climate. While these processes could be studied in the field,scale modelling can overcome many limitations of field studies and provide much greatercontrol.The central objective of this thesis is to increase understanding of the energetics andcooling in urban parks. Research on the park influence is conducted in two cities withdifferent summer climates: Sacramento, California (hot summer Mediterranean4)andVancouver, British Columbia (cool summer Mediterranean5).While this represents onlya subset of the macroscale climates where parks are important in cooling, the choice ofthese two cities will increase understanding of the range of influence of parks in differentclimates. The research adopts an integrated approach incorporating field studies thatmeasure the park effect on the surface and air thermal regimes; direct measurement ofpark SEB; and scale modelling to determine the contribution of different processes to4Köppen climate classification Csa.5K6ppen climate classification Csb.20nocturnal cooling in parks. This results in practical guidelines for urban parks designedto gain the maximum climatic benefit.1.4 ObjectivesThe specific objectives of this thesis are to:1. increase our knowledge of the park effect on urban thermal regimes by• assessing the thermal effect of parks in two cities with different summer climates;• determining the diurnal pattern of the park effect on both air and surfacetemperatures.2. measure the surface energy balance of urban parks in two cities with differentsummer climates;3. determine the relative contribution of meteorological processes to nocturnal coolingin urban parks;4. provide practical guidelines for planners regarding the manipulation of urban parksto attain maximum climate benefit.This research incorporates three distinct parts. Chapter 2 realizes the first objectiveby providing results from summertime park surveys in hot Sacramento and cooler Vancouver. Chapters 3, 4 and 5 present the methodolgy and results from park surface energybalance studies in these cities. The final part involves scale modelling of nocturnal cooling in urban parks (Chapter 6 and Chapter 7). The results from each part are combinedin Chapter 8. These are discussed and practical guidelines for planners regarding parkdesign are developed.21Chapter 2Survey of the Park Effect in Two Cities with Different Summer ClimatesThe objective of this part of the research is to provide a descriptive survey of the summertime thermal regime associated with urban parks in two cities from different climates:hot and dry Sacramento, California, and temperate Vancouver, British Columbia.2.1 Method and Instrumentation2.1.1 Survey areasSacramentoSacramento (38°39’N, 121°30’W) is located in the extensive Sacramento Valley at theconfluence of the American and Sacramento Rivers. The metropolitan area has a population of 1.48 million. Land—use surrounding the city is mainly intensive agriculturewhich requires heavy irrigation. The region has a Mediterranean—type climate with ahot dry summer. Typically the synoptic setting involves an anticyclone located west ofthe coast of California and predominantly westerly winds.Air temperature surveys were conducted for ten parks in the Mission Oaks andCarmichael suburbs of Sacramento (Fig. 2.1). Their size ranges from 2 to 15 ha. Four aremulti—use savannah—type parks with some playing fields, treed grassland, swimming poolsand childrens’ playgrounds. Two are open grassed parks with heavy irrigation; one is aforested Botanical Garden; one is a golf course and the remaining two are undeveloped,dry, savannah parks.22Figure 2.1: Location of the fixed sites in Sacramento, California, and the traverse routefor the park survey. The urban heat island traverse was from the dry rural site todowntown Sacramento.KeytoSacramento Parks1. Howe Community Park2. Cottage Park3. Orville Wright Park4. Carmichael Park5. Sufter Park6. Botanical Garden7. Jan Park8. Mission Park9. Eastern Oak Park10. Del Paso Country ClubLegendEl Irrigated greenspace——Traverse—Interstate HighwayCity RoadCity LimitsAmerican RiverScale: -0 2 kijometres23VancouverVancouver (49°15’N,123°18’W) is located in the lower Fraser Valley on the west coast ofBritish Columbia. The city lies north of the Fraser River and its Delta, and is boundedto the north by Coastal Mountains reaching 1500 m.The general climatology for the Vancouver region is presented in Oke and Hay, (1994).The region has a Mediterranean—type climate with a cool summer. The large scale flow ispredominantly from the west. In summer there are persistent high pressure systems withoccasional weak frontal disturbances arriving from the north. Local climate variationsare produced by the orography, and the proximity to water which generates land/seabreezes giving westerly winds by day and weaker easterly flow at night.Air temperature surveys were conducted for ten parks on Vancouver’s west side(Fig. 2.2). These include grassed parks (both irrigated and non—irrigated), grassed parkswith substantial shade tree borders, savannah, garden, and multi—use parks. Park sizesrange from 3 to 53 ha.2.1.2 Data collectionMobile surveys were used to collect air temperature data in the vicinity of parks. Mostsurveys were restricted to roads around the perimeter of parks (Howe and CarmichaelParks in Sacramento and Queen Elizabeth Park in Vancouver are the exceptions withroad access.) Air temperatures were measured using an automobile equipped with acontinuously—aspirated dry—bulb thermistor probe mounted in a radiation shield. Samples were taken at 0.2 to 0.5 Hz and the data continuously recorded either by a CampbellScientific (CSI) 21X data logger, or a Fluke recorder which produces a temperature trace.241j. Vancouver-A a - . .{ / <.. ‘ I Burraby /‘ Key-- tVancouver Parks‘‘\ 1. CarnarvonL / 2. TrafalgarLegend 3. Prince of WalesA. Study site . P i h m o n d 4. DevonshireUrban area 5. Van Dusen Gardens—Major road D e I ta 6. Montgomery—Highway site 7. TisdallMunicipal Boundary 8. Queen ElizabethScale: I _—.-- -- 9. Douglaso 5kilomefres 10. Connaught__—ig 52_______——King Edwa.d Ave.-‘_Legend —‘___E] Greenspace —‘—‘--— Traverse 4 1 t Ave.— Major RoadMinor Road IScale:2kmFigure 2.2: Location of the fixed sites in Vancouver, British Columbia and the traverseroute for the park survey. The urban heat island traverse was from the rural site todowntown Vancouver.25In each city the traverses consist of two main routes. One concentrated solely onthe parks, while the other was a more general traverse to measure the urban heat island(UHI) (see Figures 2.1 and 2.2). The UHI traverses were included to compare the parkcooling with that at rural locations. Traverses usually began an hour after sunset becausethe time of the maximum urban heat island in mid-latitude temperate cities is oftenbetween sunset and midnight (Oke, 1982). Traverses were conducted mainly on calmand clear nights because these conditions permit the maximum microclimate temperaturedifferences to develop. Wind speed and direction, and cloud conditions were obtainedfrom climate observations and/or the local weather service. Cloud is classified accordingto height (high, mid or low) and cover: clear (< 0.1 sky cover); scattered (0.1—0.5 skycover); broken (0.6—0.9 sky cover); or overcast (> 0.9 sky cover).Continuous measurements from fixed stations were also used to compare air temperatures between parks, the city and the surrounding rural area. These measurementsformed part of surface energy balance studies in different land—uses (Chapters 4 and 5).The Vancouver study was more detailed, and included additional surveys using bicycles, because they are not restricted to the road network. These traverses were made at4—5 hour intervals around Trafalgar Park and nearby Prince of Wales Park (see Fig. 2.2),for the period July 24—31 and August 24—25, 1992. Measurements of air and surfacetemperatures were gathered from a network of pre—assigned sites. Air temperature wasmeasured with a thermistor probe mounted in a basket on the front of the bike and surface temperatures with a hand—held Minolta infra—red Compac 3 radiation thermometer.Data were recorded manually.In addition to the ground surveys in Vancouver, surface temperatures of five parkswere remotely sensed by an AGEMA Thermovision 880 system mounted in a helicopter.This system comprises an infra—red scanner (sensitive to radiation in the 8 to 14 micrometre waveband) with a 12° field of view and a computer with software for thermal image26analysis. The thermal range of the scanner is —30°C to +1300°C and the sensitivity oftemperature is ± 0.05° C at +30°C. Flights were conducted at a height of 1—2 km, whichgave a pixel resolution of 1.5—3x 1.5—3 m on the ground.2.1.3 Data analysisAir temperatures were initially corrected to compensate for any warming or coolingduring the time of the traverse. This method assumes that the warming or cooling rateis both linear and the same at all points on the traverse, regardless of land—use (ie.commercial, residential rural). Oke and East (1971) show that this is not strictly valid,therefore the corrections are oniy approximate. Temperature spikes at intersections andother anomalies were removed. All times given are in local apparent solar time (LAT)(Oke, 1987, p.340).The maximum “park cool island” (PCI) is calculated for each traverse by subtractingthe minimum park temperature (Tn) from the maximum neighbourhood temperature(T,j:PCI=T—T. (2.1)In all the traverses the maximum neighbourhood temperature occurs in commercial land—use zones. This elevates estimates of the PCI which would be lower if a residentialmaximum is used. For traverses running into the surrounding rural region the maximumurban heat island (UHI) is also calculated:UHI=T—T,. (2.2)where T is the maximum urban temperature and Tr is the minimum rural temperature.Temperature cross—sections were plotted to assist interpretaton of spatial patterns.Isotherm maps were drawn for selected traverses to show detailed spatial patterns of27temperature. In the more intensive Vancouver studies, the diurnal variation of the PCIfor each park was also analysed.Atmospheric corrections were made to the AGEMA surface temperatures, followingthe method of Voogt (pers. comm.). This uses an atmospheric radiation transfer modelto estimate atmospheric emission and transmission using a description of the atmosphereand a model of the sensor—detected radiance which combines the atmospheric, surface,and reflected sky radiance components. Characteristics of the atmosphere were determined from vertical remote soundings, radiosonde ascents and climatological data (Voogt,pci’s. comm.). Once corrected, the images were processed by a GIS package (Idrisi) toobtain visual output.2.2 Survey results2.2.1 Spatial and temporal patterns of park influence on air temperaturesin SacramentoMobile surveysThe thermal climate of several Sacramento parks was surveyed on 6 occasions over athree—day period with variable weather in August 1993. The PCIs were better developedon August 16 and 17 which were less cloudy and less windy (Table 2.1). In this warmclimate, parks can have a substantial cooling influence (5—6° C). A pilot survey of the parkeffect in hot, dry Tucson, Arizona, measured a maximum PCI of 6.8°C (Appendix A).The air temperature trace for the traverse at 2100 LAT on August 16 clearly showsthe coolest parks are not only well—irrigated multi—use parks and a golf course, butalso dry savannah parks (Fig. 2.3). Sutter Park in particular, which is set in a semirural neighbourhood, develops a large PCI which continues throughout the night untilsunrise. This whole neighbourhood, due to its low density and semi—rural character, is28considerably cooler than surrounding suburban areas.Although the small number of traverses does not constitute a full diurnal cycle, itappears that PCI may be largest soon after sunset. At this time several park types (drysavannah, forest and golf course) exhibit cool islands greater then 5°C. In the afternoonall park types have a small to moderate PCI (1.3—2.7°C). The larger PCIs are in multi—use parks where trees lower air temperatures (through shade and possibly evaporativecooling). Soon after sunset, cooling is well established in all the parks. The maximumcooling is in the dry savannah—type where a PCI of 6.5°C was observed. Although theseparks are classified as “savannah” the trees are more dispersed than in the well—irrigatedsavannah areas of multi—use parks. The PCI in these open, dry parks is still large nearsunrise. The other open parks (grass, golf course) also have high PCI near sunrise. Theaverage PCI may be underestimated for the grass parks because one of the parks surveyedis located about 150 m away from the traverse road.Table 2.1: The average PCIs for different park types in Sacramento in August, 1993.Four multi—use types, two irrigated grass, two dry savannah, one forest park and one golfcourse were surveyed. All PCIs and UHI are in Celsius. Sunrise was about 520 LAT andsunset about 1835 LAT.Date LAT Grass Savannah Multi— Forest Golf UHI Wind Spd. Cloud(h) (irrig.) (dry) use Course (m s’) coverAug 14 1400 1.9 1.3 1.8 1.0 2.4 —3.5 6.2 high sct2100 1.1 1.0 1.0 1.0 1.0 7.2 high brknAug 16 400 3.0 3.8 2.9 3.8 3.0 2.1 clear1500 2.4 1.5 2.7 1.3 2.3 3.6 high sct2100 4.1 5.4 3.8 5.3 5.1 7.0 2.1 clearAug 17 400 2.2 3.1 1.7 3.0 2.3 3.1 high sct29C-)00)00)0EFigure 2.3: Air temperature traverse including several parks and open spaces in Sacramento at 2100 LAT on August 16, 1993. Sunset was at 1835 LAT.Legend:1. Howe Pk. (multi-use)2. Cottage Pk. (multi-use)3. Orville Wright Pk. (grass)4. Carmichael Pk. (multi-use)5. Sutter Pk. (dry savannah)6. Botanical Garden (forest) river7. Jan Pk. (dry savannah)8. Mission Pk. (multi-use)9. Eastern Pk. (grass)10. Del Paso Country Club (golf course)25242322201821190 10 20 30 40 50 60 70 80 90Time elapsed (mm)30PCI and UHI at fixed sitesThe two—day average PCI and UHI for fixed Sacramento sites is shown in Figure 2.4a (forlocations of sites see Fig. 2.1). These air temperature differentials should be interpretedwith caution since the measurements are at different heights (1.3 m park and wet rural(WR); 1.8 m dry rural (DR) and 9 m suburban (SU)). Over these two days there wassome high cloud, therefore these results do not show a maximum effect.In the early morning Orville Wright Park is only slightly cooler than the suburbansite. Soon after midday the park is actually warmer than the suburban site and remainsso until midnight. The Sacramento UHI may be negative by day. Comparison of thedifferent UHI depending on the rural reference (SU—WR, SU—DR) serve to illustrate theimportance of the character of the rural site for UHI measurements. These temperaturedifferentials depend on more than the “urban” or “park” environment of interest — thenature of the base environment must be known and stable.The cooling after sunset (Fig. 2.4b) shows the park rate is similar to that at the dryrural site. The dry rural site initially cools more quickly but for the latter half of thenight these two sites have a similar rate. The wet rural site cools more slowly, at a ratesimilar to suburban site. Despite the rapid cooling of the park, the nocturnal PCI issmall. This is because the park is about 2°C warmer than the suburban site in the lateafternoon and park cooling must first erode this deficit before a PCI develops.Summary of the park effect in SacramentoIn Sacramento, a dry savannah park in a semi—rural setting had the largest observedPCI. Irrigated parks can also create substantial cool islands, similar in magnitude tothose observed in a pilot survey in Tucson. The irrigated and well—treed forest park andgolf course induce the greatest effect (about 5°C). The UHI observed by mobile traverses31C)C00C)a) PCI and UHI from fixed sites—1—2—3LATb) Nocturnal cooling from fixed sites0—2—4—6—8—10—12Figure 2.4: The PCI and UHI from fixed sites in Sacramento. The two—day averagePCI, and UHI based on suburban—dry rural difference (SU—DR) and suburban—wet ruraldifference (SU—WR) are shown in (a) while the nocturnal cooling after sunset is shownin (b).Time after sunset (h)32show the city to be a cool island in the afternoon, compared to dry rural areas, but aftersunset it ian be as large as 7°C. However, as windy conditions prevail at Sacramento,the creation of large PCIs and URIs is relatively rare.2.2.2 Spatial and temporal patterns of park influences on surface and airtemperature in VancouverOverviewNineteen automobile traverses were conducted in the vicinity of parks in Vancouver inthe summer of 1992. There were three distinct study periods — June 22/23, July 28/29and the most intensive — August 24/25. The weather for most traverses was mainly clearand calm; high scattered cloud occurred on a few occasions.There is considerable temporal and spatial variability in the mean PCI (Fig. 2.5).It has a peak soon after sunset (about 3°C) which corresponds with a decrease in windspeed. The mean PCI then declines through the night but then increases again beforesunrise. In the late afternoon when wind speed is at a maximum, the mean PCI is only1.4°C. The extreme high PCI (about 5°C) occurs in garden parks soon after sunset.Shortly before sunrise open, grassed parks have a high PCI near 4°C. Another extremePCI occurs in the afternoon in grass parks with large tree borders (about 4.5°C).Tkaverse analysis for intensive period — August 24/25, 1992This discussion reports observations from three surveys: park surface temperatures measured by the AGEMA system mounted in a helicopter; and air temperatures measuredfrom automobile traverses and bicycle traverses.334530E-o40(01•I I11 14 15 17 19 21 23 1 3 5 7LATFigure 2.5: PCI range and mean for parks surveyed in Vancouver in summer, 1992. Thedata are averages from 3 survey periods with comparable weather. The average windspeed is also shown as a dashed line.Surface temperatureTo assist interpretation of surface temperature patterns, aerial photographs of these parksare given in Figure 2.6. Four of the parks are grassed, but have very different soil moistureregimes. Tisdall and Prince of Wales Parks both have sand—based playing fields whichare normally irrigated, maintaining a soil moisture of about 30%. Trafalgar Park wasirrigated until a sprinkling ban in June, and by late August the soil moisture had droppedto about 13%. Montgomery Park and the field adjacent to Prince of Wales Park, arevery dry (soil moisture 9%) because neither was irrigated.Surface temperature images for the four flights (early afternoon, shortly after sunset,midnight and around sunrise) are given in Figures 2.7 to 2.10. In the early afternoonthe surface temperature range is the greatest; tree canopies have temperatures of about18°C whereas hot roofs approach 50°C. The dry grass parks have surface temperatures34comparable to built surfaces (35—37°C), while the irrigated grass parks are cooler (24—29°C). This is clearly seen in Figure 2.7c where irrigated and dry fields are side by side. Incontrast, in the well—treed Shaughnessy area, surface temperatures are cooler throughout,and it is harder to locate Devonshire Park in the thermal image.Shortly after sunset the parks establish cool islands. The drier open grass parks havethe coolest surface temperatures (<8°C) while the irrigated fields are slightly warmer(9—12°C). The well—treed suburb is much warmer (12—14°C) with only the clearings inDevonshire Park showing substantial cooling.By midnight the very dry Montgomery Park has the lowest temperatures (3.5—6°C).The other dry parks such as the field near Prince of Wales, and Trafalgar parks, havetemperatures less than 8°C. Irrigated playing fields are much warmer, particularly atPrince of Wales Park where irrigation was occurring. At Shaughnessy the temperaturesare uniformly warm with small patches (gardens and clearings) slightly cooler.Near sunrise the patterns are the same as observed through the night, however thetemperature contrast between the dry and moist parks has further increased.Automobile traverse air temperatureAugust 24 and 25 were good days for thermal differences to develop. In mid afternoonof the 24th the UHI was about 4°C, increasing to 5.9°C near sunset with a peak nearmidnight at 9.8°C, and remaining around this value until sunrise. This is within 0.4°Cof the maximum ever observed in Vancouver (Oke, 1981). Throughout the day, thermaleffects of parks are clearly identifiable. Parks are represented by marked dips in thetemperature trace, a “cool island”, relative to their neighbourhood (Fig. 2.11). Theintensity of the PCI appears to depend upon the time of day and type of park (Fig. 2.12).Whilst the diurnal patterns are not markedly different, certain consistent featuresrelate to park type. Open grass parks have a relatively uniform mean PCI of about 2°C35(a) Montgomery Park (dry, grass) (b) Trafalgar Park (moist, grass)(c) Prince of Wales Park (irrigated grass)(e) Devonshire Park (moist, savannah)Figure 2.6: Aerial photographs of Vancouver parks used in helicopter survey of surfacetemperatures.(d) Tisdall Park (irrigated grass)*1aaaaaaaç:(c) Prince of Wales Park (irrigated gras&) (d) Tisclall Park (irrigated grass)(e) Devonshire Park (moist, savannah)r/ aIciris±Figure 2.7: Surface temperatures of Vancouver parks and neighbourhoods at 1400 LATon August 24, 1992.(a) Montgomery Park (dry, grass) (b) Trafalgar Park (moist, grass)37(a) Montgomery Park (dry, grass) (b) Trafalgar Park (moist, grass)714(c) Prince of Wales Park (irrigated grass) (d) Tisdall Park (irrigated grass)(e) Devonshire Park (moist, savannah)Figure 2.8: Surface temperatures of Vancouver parks and neighbourhoods at 1930 LATon August 24, 1992. Sunset was at 1919 LAT.38(a) Montgomery Park (dry, grass) (h) Trafalgar Park (moist, grass)(c) Prince of Wales Park (irrigated grass) (d) Tisdall Park (irrigated grass)4Figure 2.9: Surface temperatures of Vancouver parks and neighbourhoods at 2300 LATon August 24, 1992.(e) Devonshire Park (moist, savannah)39(a) Montgomery Park (dry, grass) (b) Tra.falgar Park (moist, grass)—(c) Prince of Wales Park (irrigated grass) (d) Tisdall Park (irrigated grass)icLrii(e) Devonshire Park (moist, savannah)Figure 2.10: Surface temperatures of Vancouver parks and neighbourhoods at 500 LATon August 25, 1992. Sunrise was at 518 LAT.40C)0ci)-C)ci)ciE‘a)Figure 2.11: Trace of air temperature from a traverse inVancouver at 1900 LPkT OflAugust 24, 1992. Sunset wasat 1919 LAT.Legend:1. Trafalgar Pk. (grass)2. Prince of Wales Pk (grass, sand-based)3. Devonshire Pk. (savannah)4. Van Dusen Gardens5. Montgomery Plc. (dry grass)6. Tisdall Pk. (grass, sand-based)7. Queen Elizabeth Pk (multi-use)8. Douglas Pk. (grass, tree border)9. Connaught Pk. (grass, tree border)10. Camavon Pk. (grass)1817161514130 10 2030 40 50Time elapsed (mm)6041a) Gross parks b) Gross parks with tree borderFigure 2.12: Average PCIs for different park types in Vancouver on August 24/25, 1992.The survey comprised five grass parks (with varying soil moisture), two grass parks withlarge tree borders, a garden and a multi—use park and one savannah park.432// / I Ii43207/IpI /17/////C)()a00a-I7/// IiI I7//////400a0C)a-f f f -r 111 14 15 17 19 21 23 1 3 5 7LATc) Garden and multi—use parks03. 02:Iiaiii43‘, -f-’ t- -r -1 -1 -r -11 14 15 17 19 21 23 1 3 5 7LATd) Savannah parkLi.•i Ii2LAT11 14 15 17 19 21 23 1 3 5 7 11 14 15 17 19 21 23 1 3 5LAT742through the day (Fig. 2.12a). The mean PCI increases soon after sunset and reaches amaximum close to 3.5°C shortly before sunrise. Grass parks with tree borders have amaximum effect in the afternoon, with a secondary peak near sunrise (Fig. 2.12b). Thegarden, multi—use and savannah parks show a distinct and large (4°C) mean PCI soonafter sunset (Fig. 2.12c and d). They have a similar pattern of a decrease in the PCIbeyond this maximum, and then a gradual increase towards sunrise.Bicycle traverse air temperatureFigure 2.13 demonstrates the microscale variability of air temperatures around TrafalgarPark and Prince of Wales Park. The traverse at 1400 LAT shows little difference in airtemperature between the parks and the neighbourhood. However, after sunset the cooling effect of the park quickly becomes established. The air in the vicinity of both parksis 1—2° C cooler than their surrounding environment. By 2300 LAT these temperaturedifferences are accentuated, and Trafalgar Park has a PCI of approximately 2.5° C. Comparison of air temperatures with surface temperatures for Trafalgar Park at this time(see Fig. 2.9b) shows that cooler air has been advected beyond park boundaries, with thepark’s influence felt up to about the park’s width away. The coolness of the surroundingneighbourhood may be due in part to cold air ponding due to drainage from higher land,but Trafalgar Park is clearly evident as a “cold island”. Prince of Wales Park is warmerthan Trafalgar, probably due to the fact it is at a higher elevation. Shortly before sunrisethe park is only about 0.5°C cooler than the urban surrounds’.This more detailed survey confirms that the cooling effect for open parks is mostpronounced after sunset, and that it extends beyond the boundaries of the park. Bicycle traverses have the advantage of measuring air temperatures within parks to get a‘This neighbourhood is mainly residential. This explains the lower estimates for park cooling compared to Figure 2.12 which uses the maximum suburban temperature to derive the PCI.43‘1 .(tt0I-icsq -1)-(tO I—(t(DCLU jjb iiiriLl 1W IllECEIE1EWill ww m LUDU WI LUI.ecLULUÜLJWWWmore accurate measure of the PCI. A comparison between air temperatures measuredby bicycle and those measured from the mobile automobile traverses, which only measure air temperatures around the perimeter of the park, suggests that the latter mayunderestimate the nocturnal PCI in these parks by up to 1.5°C. On the other hand, byday automobile traverses may overestimate the PCI, particularly for drier parks wherethe surface can be warmer than near the bordering tree canopy where the automobilesamples.PCI and UHI from fixed sites.Continuous measurements of air temperatures at fixed sites in an open grass park (Trafalgar), suburban (Sunset), downtown and rural (Delta) areas (see Fig. 2.2), allows comparison of PCI and UHI development together with their nocturnal cooling characteristics(Fig. 2.14a and b). As measurements were taken at different heights: 1—1.5 m at thepark and rural sites; 11 m at the suburban site; and 3 m at the downtown site, the trendsshould be interpreted with caution.Trafalgar Park is slightly warmer than the suburban site by day but is about 2—2.5° Ccooler at night (Fig. 2.14a). The UHI (suburban—rural difference here) is larger at night;close to 5°C. Throughout the night, the park and the rural site have very similar coolingrates (Fig. 2.14b) which are considerably greater than for the city sites, especially inthe first two hours after sunset. This divergence is probably due to differences in viewfactors (greater sky view factors at open sites enhance radiative cooling) and possibly inthermal admittance. Natural surfaces often have lower thermal admittance which meansthey have little heat to release and this heat is released more readily. This establishesthe maximum difference in temperature between the land—uses, within three hours aftersunset. The cooling rate in the suburban area remains almost constant throughout thenight. At the downtown site the rate is more variable.45The park cooling is comparable to the rural case but the nocturnal PCI is less thanthe UHI. This phenomenon was also observed in Sacramento. It occurs because the parkis warmer than the suburban site in the late afternoon. Thus park cooling must firsterode this deficit allowing only a small PCI to develop.Summary of the park effect in VancouverThe surveys in Vancouver were all conducted under “ideal” conditions (calm and clearskies). Several researchers have observed maximum surface temperature contrasts by dayin urban areas with a reduced range at night (the opposite to air temperature patterns).The Vancouver survey confirms this: surface temperature contrasts between greenspaceand the built environment peak in the early afternoon. However, open dry grass parksare not necessarily cooler than urban areas. Nocturnal cooling is most pronounced inparks with high sky view factors and dry soils. A savannah park is much warmer bynight probably because the canopy reduces radiative loss.The ensemble average PCIs for all parks surveyed show considerable temporal andspatial variability. However, there are consistent patterns relating to park type. Parkswith substantial tree canopies have maximum PCI in the afternoon. Garden, multi—useand savannah parks peak soon after sunset, while open grass parks have a maximum PCInear sunrise.2.3 Comparison of PCIs in cities from hot and cool summer MediterraneanclimatesThe air temperature surveys in the two cities confirm and extend earlier findings of thethermal effects of parks. Observations in Vancouver are fairly consistent with Oke’s(1989) discussion that the park effect in mid—latitude cities is typically 1—2° C and rarely46(a) PCI and UHI2a°Sunset0 500 1000 1500 2000LATb) Nocturnal cooling from fixed sites0 1 2 3 4 5 6 7 8 9 10Time after sunset (h)Figure 2.14: PCI and UHI, and nocturnal cooling for fixed sites in Vancouver. The dataare four—day averages (July 25—28, 1992). The PCI is a suburban—park air temperaturedifference and the UHI is a suburban—rural air temperature difference (a). Nocturnalcooling for different land—use types is shown in (b).47more than 3°C. The park effect in Vancouver is slightly greater than previous estimates.Under ideal conditions PCIs approach 5°C.The PCI and UHI in Sacramento and Tucson show many similarities. In both citiesthe PCI for irrigated parks can be substantial (5—7°C). Several of the observations werenot conducted under ideal conditions. This suggests a larger effect may be possible. Whilethe PCI is typically larger than in mid—latitude temperate cities, the UHI is somewhatsmaller (2.5—7°C). By day, the city may be a cool island. Within this cool island are evencooler parks.In each city the largest PCI are generally found in larger multi—use and garden parks,and golf courses. These parks have a PCI that is well established in the afternoon.Soon after sunset there is a marked increase in the urban—park temperature differential.Well—treed parks often have their maximum PCI at this time. Irrigated open grass parksusually have higher PCIs near sunrise, while unirrigated ones cool rapidly after sunsetand continue to cool at a greater rate than other park types, establishing their highestPCIs in the hours prior to sunrise. Despite the large PCIs observed in all cities, the parkeffect remains fairly localized: up to about the park’s width away.These observations suggest that during the day trees may play an important role inestablishing a cool park effect, perhaps through a combination of shade and evaporativecooling. At night it appears that the surface geometry and moisture status of the parkare important controls on surface cooling. Open parks (with higher sky view factors) thathave dry soils (and hence lower thermal admittance) cool the most. The contribution ofthese processes to nocturnal cooling is investigated through scale modelling, in Chapters6 and 7.The potential park effect is largely determined through macroscale climate and thecharacter of the urban surface. For a given park type, in a given urban setting (e.g.suburban) the potential for it to be cooler than its urban surrounds depends on the48macroscale climate. For example if the Van Dusen Gardens in Vancouver were transplanted to Sacramento, the effect of this park would be increased. The warmer theclimate, and the drier the climate, the greater the effect.49Chapter 3Conceptual Framework for Field MeasurementThis chapter discusses the framework for surface energy balance (SEB) measurements inurban parks. First there is a discussion on the influence of the advective regime on thespatial variability of the SEB. This considers spatial variability in response to fetch (thedistance of airflow in the upwind direction) and the turbulent regime in a park. Thesecond part of the chapter discusses the basis for the selection of site locations. Thechapter concludes with an overview of techniques used to measure the SEB fluxes.3.1 Spatial variability of SEB across a parkThe SEB framework for an open grassed park is more similar to the rural case of apasture or low crop, than the urban case. With a low grass canopy, the storage termis replaced by the gound heat flux, QG, and the net biochemical heat storage, Qp, issufficiently small to be neglected. However, when placed in an urban setting a park islikely to experience advective effects so the advection term must be included. Thus theenergy balance for a park is given by:Q*Q+Q+Q+Q (3.1)3.1.1 Influence of fetch on the SEB in a parkRather than taking measurements at a central point in the park, several sites are chosento provide an indication of the spatial variability of the surface energy balance. Net50radiation and soil heat flux are likely to be spatially relatively conservative in an openpark (providing there are no trees and the soil is homogeneous), but the turbulent fluxesare likely to vary. For example, the latent heat flux varies in response to fetch (distancefrom the boundary between the park and the comparatively warm and dry suburbanarea). This is the “leading edge”or “fetch” effect (Oke, 1987) whereby QE is greatestat the leading edge (i.e. the park boundary) because of microscale advection of dry air,and decreases rapidly with distance into the park. If the park is moist compared to thesurrounding neighbourhood, and large enough, it may also act as an “oasis” (Tanner,1957). In this case the advection is accompanied by subsidence of warm air from theupwind neighbourhood, over the cool park which supplements the radiative energy supplyand thereby makes it possible to sustain abnormally high rates of evapotranspiration.In the absence of advection, evapotranspiration from a well—watered, extensive surface, has a lower limit termed “equilibrium evaporation”, by Slatyer and Mcllroy,(1961):Qq (Q*- QG) (3.2)where s is the slope of the saturation vapour pressure—temperature curve (0.145 k Pa0 C’at 20°C) and-y is the psychrometric constant (-y = cp/(Le) where c, is the specificheat of air at constant pressure (1.01 J g°C), L is the latent heat of vaporization(2440 J g’ at 20° C), p is the density of air (1.20 kg m3 at 20° C) and is the ratio ofthe molecular weight of water to that of dry air (0.622)). The value of s/(s +7) at 20°Cis 0.69.Priestley and Taylor (1972) used equilibrium evaporation as the basis for estimating“potential evapotranspiration”, QE. They defined this as the evaporation that wouldoccur from a large—scale, well—watered region in the absence of advection. It is estimated51by:QEp = crQEq (3.3)where is an empirical constant. They showed that for oceans or extensive wet vegetatedsurfaces, evaporation was about 26% greater than (i.e. c = 1.26). Equation 3.3 with= 1.26 has had widespread application and gives good estimates of evaporation evenfor areas much smaller than “regions”.Brutsaert (1982) suggests that over wet surfaces, equilibrium conditions are encountered only rarely, if ever. This is because the atmospheric boundary layer is never atruly homogeneous boundary layer, but is continually responding to large scale weatherpatterns. This contention was supported by = 1.26 to 1.28 observed over oceans orlarge wet surfaces. Hence he suggests that evaporation from these surfaces is assistedby large—scale advection. Similarly, McNaughton and Spriggs (1989) suggest that whilePriestley and Taylor’s potential evaporation is valid for estimates of evaporation in theabsence of local advection, QEp may include some large—scale advection. This is becausePriestley and Taylor did not consider energy exchanges between the planetary boundarylayer and the atmosphere at large. The additional evaporation comes from the entrainment of warm, dry air from the inversion above the growing daytime planetary boundarylayer.Lang et al. (1974) suggest that the advective component of evaporation, QEadv canbe determined from:QEadv = ax = QE — QEp (3.4)where a and p are empirical constants and x is the distance downwind from the leadingedge. Parameter a is the amplitude of the advective evapotranspiration and p describesits rate of decay downwind. Exponent p is obtained from the profile law for windspeed1The relation was developed for regions measuring tens of thousands of square kilometres.52and Lang et al. (1974) estimate it to be 1/6 for crops. This relation implies rapiddecay in evaporation immediately downwind of the leading edge, and a gradual decreasetowards potential evaporation with increasing distance. Equation 3.4 may underestimatethe advective component by using QE as a base reference to determine advective effects.Few researchers distinguish between edge— and oasis—effects i.e. microscale and mesoscaleadvection. Often the confounding effects of proximity to a leading edge are avoided bysiting instrumentation within an extensive homogeneous area. However, in the contextof an urban park it is useful to make the distinction between edge and oasis effects. Aconceptual framework for distinguishing these effects is developed after the method ofBrakke et al. (1978).Brakke et al. (1978) inferred advective effects by comparing evaporation with theterm Q*— QG. At a considerable distance downwind from the leading edge QE does notchange with distance. They suggested that the elevation of QE over Q* — QG was dueto regional sensible heat advection. Near the leading edge QE surpasses this value andthey suggested this was due to local advection.It appears that Brakke et al. s (1978) use of Q*—QG as a base reference for potentialevaporation in the absence of advection, is too high. This framework is revised usingequilibrium evaporation which is a more accepted definition for evaporation in the absenceof advection. The advective scales are also revised on the basis of the scale classificationof the urban canopy layer (Table 1.1).It is suggested that due to the edge effect, evaporation in a wet park will show anapproximately exponential decay immediately downwind of park boundary (Fig. 3.la).The edge—effect results from microscale advection from surfaces immediately upwindfrom the park boundary. With distance into the park, evaporation will decrease until,if the park is large enough, it becomes steady (i.e. unchanging with distance). At thisdistance, the elevation of evaporation above the equilibrium rate indicates the degree53a) microscale advection b)ttQE QE QE Q,mesoscaleadvection / mesoscaleQ ______jL..... Qqurban park urban parkDistance —* Distance —*Figure 3.1: Postulated variation of the latent flux with distance of fetch for an irrigated(a) and (b) an unirrigated park.of mesoscale advection. It is foreseeable that in urban areas, oasis—type advection mayfurther enhance evaporation. This is an extreme case of mesoscale advection and is saidto occur if QE> QEp.Several studies of evapotranspiration from irrigated fields surrounded by semi—aridareas report both edge and oasis effects (see Rosenberg et at., 1983 for a review). Forexample, a number of observational studies at Davis, California (near Sacramento) (eg.Dyer and Crawford, 1965; Goltz and Pruitt, 1970) used a series of air temperature profilesacross irrigated grass fields set in the midst of dry fields to infer QH. The latent heat fluxwas estimated from both closure of the energy balance (Q* and QG were measured) andlysimetry. Lang et at. (1974) used lysimetry to study the influence of local advection onevaporation from an irrigated rice field and Rider et al. (1963) measured evaporation from54irrigated grass downwind from a paved surface. These studies all show local advectionconsistent with Fig. 3.la. With increasing fetch the edge effect decays until, finally,it adjusts fully to the new surface and remains approximately constant with distance.Estimates of the extent of the edge effect vary widely from less than 20 m (Rider et al.,1963) up to 200 m (Rijks, 1971). Local advection contributes to increased evaporation inurban areas. Oke (1979) found advectively—assisted evapotranspiration over an irrigatedsuburban lawn in Vancouver, B.C.The oasis effect has been reported from observations over crops at many locations(e.g., Fritschen and van Bavel, 1962; Rosenberg, 1969 and 1972; Wright and Jensen,1972; Blad and Rosenberg, 1974; Rosenberg and Verma, 1978). Even at considerablefetch these studies show evaporation is significantly higher than the potential rate. Forexample, Rijks (1971), in a study of water use by irrigated cotton in the Sudan, foundevaporation rates 1.8 times greater than the supply of net radiation when both edge andoasis—type advection were important. The oasis effect alone elevated QE/Q* to 1.5.In an unirrigated park, the spatial pattern of QE may be very different (Fig. 3.lb). Itis suggested that if irrigation is not applied, soil moisture content near the upwind edge islikely to become depleted in comparison with the rest of the park, due to the cumulativeloss of soil water associated with microscale advection. Eventually such enhanced QE atthe upwind edge of the park can no longer be sustained. Further away from the edge,QE is greater than that of the background suburban value. Even if the surface is notsaturated, mesoscale advection will probably elevate evaporation above the equilibriumrate. The size of this suburban—park difference in QE probably depends upon manyfactors including the regional climate, and soil moisture content, suburban density andgardening practice.553.1.2 Influence of the turbulence regime on the SEB in a parkThe turbulence regime in a parkSince a park may be surrounded by buildings and/or trees, air—flow distortion can occur,especially near the upwind boundary. In the absence of work directly on flow in parks,there are three main sources of evidence that can assist understanding: studies of airflowaround buildings, windbreaks and forest clearings. Research on airflow around buildingshas concentrated on the turbulence field upwind and downwind of the building in orderto model pollutant dispersion. Less consideration has been given to the spatial variabilityof turbulent fluxes downwind of a building, however, this work can yield important clues.Windbreak research is the most useful in providing data on the variation of turbulentfluxes behind a barrier. It often focuses on the variation of evaporation with distance fromthe barrier, so this literature is of direct interest. Finally, work on turbulence in forestclearcuts, can be of assistance. To date there has not been much quantitative researchbut some wind—tunnel studies can assist in interpretation of patterns of turbulent fluxeswithin a park.The flow field immediately downwind of an isolated building is characterized by thewake cavity, which is a zone of low pressure and reduced mean windspeed, in which flowrecirculates. Beyond the cavity is the turbulent wake zone in which a large velocity gradient interacts with a large downward momentum flux to produce considerable turbulence(Fig. 3.2). The length of the cavity (Xr) depends on the building dimensions (height,H, length in the along—wind direction, L, and width of the building in the cross—winddirection, W), the speed and vertical variation of the incident wind, the intensity andscale of the incident turbulence, the stability of the approach flow, the aerodynamcisof the upwind surface, and the fluid viscosity (Hosker, 1984). There are several studies(mostly wind—tunnel) that give quantitative data on the dimensions of the wake cavity.56Figure 3.2: Flow distortion over a building — the cavity and tubulent wake zones. (FromHosker, 1984 based on Woo et zL 1977 and Hunt et al. 1978.)Hosker (1986) in a review of the literature, reports a study by Fackrell and Pearce thatestimates the along—wind extent of the cavity of an isolated building to be:1 QW(35)H—(°(i + 0.24))Typically, for approach flow perpendicular to a building with a W/H of 1, the quietzone extends to about 5 H and increases to about 12 H as W/H increases to 40 (Hosker,1984). If the building is oriented at an angle to the approach flow, the cavity is considerably broadened and lengthened (Robins and Castro, 1977; Ruck, 1993). The horizontalextent of the turbulent wake zone depends on the dimensions and orientation of thebuilding, but Hosker (1984) suggests that the wake typically extends at least 10 to 20 Hdownwind. The turbulent wake zone increases if the wind blows at an angle to thebuilding and in stable conditions.Windbreak studies typically assess impacts on the microclimate in the lee of theobstacle. The characteristic flow around a windbreak is similar to that around a buildingLINF57with both a cavity and a turbulent wake zone. Raine and Stevenson (1977) termed thiscavity the “quiet zone”, reflecting the reduced wind speed and turbulence dominated bysmall eddies. However, unlike buildings, windbreaks may be porous allowing some flowto “bleed” through the shelter (Fig. 3.3) so the nature of the quiet zone depends on thenature of the barrier (solid or porous) and the turbulent characteristics of the approachflow. McNaughton (1988) reports that denser barriers produce the largest reductions invelocity close in their lee and that the quiet zone is longer behind such barriers. Onemust also consider the height of the obstacle (H) in relation to the surface roughness,(zo). At higher values of H/zo greater reductions in mean wind speed occur in the lee anda larger quiet zone is observed (van Eimern et al., 1964). Windbreak studies suggest thatthe quiet zone extends to about 8 H behind a long windbreak in near—neutral conditionswith the approach flow perpendicular to the windbreak and H/zo 100 (McNaughton,1988). As for buildings, increasing stability greatly increases the extent of the quiet zone,and instability decreases it. Little research has been done on the turbulent wake zone.The final source of data that is useful in determining the likely turbulence regime ofa park is that of turbulence research at forest edges and clearcuts. The analogy betweena forest clearcut and a park may be a useful one. Oke (1989) explored the similarities between the urban canopy layer and a forest canopy. With regards to aerodynamicproperties, he noted that buildings are true bluff bodies because of their impermeability,inflexibility and sharp edges. Although trees are good generators of mechanical turbulence, buildings are far more effective roughness elements. Despite these differences itmay be of some merit to consider studies of turbulence in forest clearcuts. The conceptof canopy—airstream coupling may be similar in both cities and forests and in turn inurban parks and forest clearcuts.As McNaughton (1989) noted, this field has received little attention. While someresearch has focused on wind regimes associated with forest edges (e.g. Raynor 1971,58__H D—_Obstacle o 4 8 12 16Quiet zone Turbulent wake Readjustmentzone zoneFigure 3.3: Flow distortion over a barrier — the quiet and turbulent wake zones. (AfterMcNaughton, 1988.)Bergen 1975, Gash 1986) few studies have concentrated on turbulence. Chen et al.,(1994) simulated a forest clearcut in a wind—tunnel and made intensive turbulence measurements. They observed a quiet zone that extended to about 3 H, a turbulent wakezone from 3 to 18 H, followed by a readjustment zone.A significant difference between a park in an urban canopy and a windbreak in aforest is the geometry imposed in the urban case by streets. It is possible that streetgeometry could channel air—flow into and out of, parks. These jets could significantlyalter the turbulence regime in the park.In conclusion, despite the absence of turbulence measurements, it may be possible toinfer the turbulent regime within a park, based on the foregoing discussion. The extentof the quiet and turbulent wake zones will depend greatly on the nature of the parkboundary — whether buildings, vegetation or both, the angle of the prevailing wind tothis boundary and on turbulent charactersitics of the approach flow.59Impacts of the turbulent regime on the SEB in a parkFrom the foregoing, if the suburban—park boundary is abrupt, we may expect threedistinct turbulent regimes to exist in the park (quiet zone, turbulent wake zone andreadjustment zone). This will bring implications for climate variables such as humidityand temperature as well as the SEB fluxes. McNaughton (1988) discussed the likelypatterns of these in both the quiet and turbulent wake zones.In the quiet zone, eddy diffusivity is decreased due to the reduction in turbulentkinetic energy and eddy size, so the turbulent transport of scalars is suppressed. Henceit is expected that both surface temperatures and humidity will be higher during the day.The wake zone is characterized by considerable turbulence, consequently it is dominatedby larger eddies giving enhanced eddy diffusivity and turbulent transport of scalars. Thisresults in lower surface temperatures and humidity. The study by Chen et at. (1994)found a pattern of increasing eddy diffusivity with distance from the forest edge, reachinga maximum at 6 H in the turbulent wake zone (Fig. 3.4).Effects on the SEB are more complex. As discussed earlier, net radiation and theground heat flux are not likely to vary greatly across an open grass park with homogeneous soil, except at the park edges if there are any shadows cast by a vegetated or builtborder. However, since both Q* and QG depend on surface temperature there is likelyto be some variability of these fluxes between the quiet zone and the turbulent wakezone. With higher surface temperatures in the quiet zone Q* may be reduced due tothe increase in out—going long—wave radiation, while QG will increase. Conversely in theturbulent wake zone, where mixing tends to smooth out the diurnal temperature wave,Q* will be higher and QG will be reduced (McNaughton, 1988).Due to suppressed turbulent transport in the quiet zone, it is expected that theturbulent fluxes QE and Qj will be reduced. However, the situation is more complex,60Cl)4 6 12 16Distance from ForestEdge (Tree Heights)Figure 3.4: Eddy difFusivity patterns in a forest clearcut (from Chen et al., 1994).Wake Zone(enhanced turbulent exchange) ReadjustmentZone00 2061due to the dependence of the evaporation rate on the saturation deficit. Evaporation, E,from a plant canopy can be described(3.6)where r is the canopy resistance and D0 is the surface value of the saturation deficit(Monteith, 1963). If the canopy resistance is uniform across a field then changes inevaporation reflect changes in D0. The surface saturation deficit will vary depending onthe degree of “coupling” between the surface and the conditions overhead. If the surfaceis well coupled, such as in the turbulent wake zone, D0 will approach the saturationdeficit overhead, D, at a level uninfluenced by the windbreak. However in the quietzone with reduced wind speed and turbulence the surface is effectively decoupled fromthe flow overhead so D0 will increase or decrease depending on the relative amounts ofheat and vapour added (McNaughton, 1988).McNaughton and Jarvis (1983) introduced a decoupling factor (ft) that varies fromo when the surface is completely coupled to conditions overhead, to 1 when the surfaceis completely decoupled. In an open field is given by:= [1+(c)f’ (3.7)where r is the canopy resistance, ra is the aerodynamic resistance and e = sL1/c.In the quiet zone it is expected that will be larger than upwind as the sheltered surface is more decoupled from conditions above. When the surface is completely decoupledequilibrium evaporation prevails given by equation 3.2. This is equivalent to viewing D0in equation 3.6 as having an “equilibrium saturation deficit” (McNaughton, 1983):f- \n —— dG)rc-‘-‘eq — pL(s+7)62Given the concept of coupling, McNaughton (1988) then suggests that a change in coupling (M) produces a change in D0:LD0 = Ml(Deq — D) (3.9)and a change in evaporation ratetiE = tifZp(Deq — D) (3.10)from the canopy—model equation 3.6. Thus evaporation in a sheltered area can be increased or decreased depending whether the air overhead is drier or wetter than Deq, andalso on the sign of zMi Under conditions of dry air advection the latent heat flux willbe reduced in the quiet zone and enhanced in the turbulent wake zone.The sensible heat flux can be considered as the residual after QE is taken from theavailable energy, so as QH increases, QE decreases. Hence under conditions of dry airadvection QH will be increased in the quiet zone and decreased in the turbulent wakezone (McNaughton, 1988).In summary, understanding the spatial variability of the SEB components in a parkrequires consideration both of the fetch and turbulent regimes of the park. Microscaleadvection is likely to occur at the upwind edge of parks if a moist park is surroundedby a warm, dry suburban neighbourhood. Thus one expects an exponential decay inevaporation with distance from the upwind park edge. However, urban air flowing fromroof—level down into the park creates a turbulence regime that has a quiet zone at theupwind edge, and a turbulent wake zone further into the park. The dimensions of thezones will vary depending on characteristics of the approach flow and on the nature of thepark boundary. In the quiet zone Q* may be slightly reduced while Q may be slightlyincreased due to higher surface temperatures. Evaporation is substantially reduced dueto decoupling of the surface from conditions overhead, which results in an increase in63windo QuietzoneLegend:Q QQQReadjustmentzoneFigure 3.5: Scenario for likely impacts of both fetch and the turbulence regime on thespatial variability of the daytime SEB in a suburban park. The length of the arrowsindicates the approximate proportions.QH. Conversely, in the turbulent wake zone increased turbulent transport reduces thedecoupling, and QE is enhanced, while QH is reduced. These concepts are illustrated inFigure Site locationsThe observation programme is designed to measure the surface energy balance of twosuburban parks with similar characteristics (size, water status and vegetative cover) indifferent summer climates, viz:• Sacramento, California — hot summer Mediterranean;• Vancouver, British Columbia — cool summer Mediterranean.Turbulentwake zone‘AUrban Park64The choice of two background climates is to assess any differences in the influence ofparks in different climatic regimes. The traverse data indicate that the park cool islandis greater in a hot dry climate. Energy balance measurements in both climatic regimesare designed to understand the driving forces of park climate. To achieve this, it isnecessary to place the park in context with respect to its surroundings. Hence, in eachstudy, observations are simultaneously taken at suburban and rural sites. A comparisonof energy balance fluxes and climatic parameters at each site helps elucidate controls onpark climate.To facilitate study and interpretation, the research is conducted in a simple park system. In this reductionist approach, the park should be an open, well—irrigated, grassedpark, in a flat area away from local topographic influences. This is to simplify measurements and reduce any confounding effects of different or layered vegetation or localdrainage flows. In addition the park should be located in a residential area. This is toenable comparison with flux measurements available concurrently from a tower in a residential suburb. Research into the park system in its simplest form provides a tractableproblem but does limit the generality of results. Finally, from a practical standpoint, thepark must be accessible and secure, because equipment is installed for up to two weekswith continuous monitoring.A compromise had to be reached regarding these criteria. In Sacramento, OrvilleWright Park was selected, but a tree border at the upwind edge subjected measurementsto greater edge effects than desired. In Vancouver, an unforeseen summer ban on irrigation of any grassed areas meant that most city parks were not irrigated unless they weregrassed, sand—based playing fields. Also, access to parks was a problem, and severelylimited the choice of park. Trafalgar Park was chosen, however it is located in a slighttopographic depression and at the upwind boundary there is a school and a small step(0.8 m) in elevation to a gravel playing field which might cause flow distortion near the65leading edge. Due to the irrigation ban the water status of the Vancouver park was lesscomparable to that of the Sacramento park than wished for.Given the expectations of spatial variability in the turbulent fluxes outlined in section 3.1, the measurement programme in each park is designed to include several sitesalong a transect aligned with the prevailing wind. Sites are spaced exponentially fromthe upwind edge to the centre of the park.3.3 An overview of techniques used to measure the SEB fluxesIn this research several techniques are employed to measure the SEB of park, suburban and rural surfaces. These techniques are outlined below; details of sampling andinstrumentation are given in Chapters 4 and 5.Eddy correlationMost micrometeorological methods to measure the turbulent fluxes require certain atmospheric conditions to be met and extensive surface homogeneity. Of the techniquesto measure both sensible and latent heat, eddy correlation is theoretically the best andhas few limiting assumptions. It provides a direct estimate of the sensible heat fluxby simultaneously measuring turbulent velocity and temperature fluctuations and determining their covariance over the desired averaging time. Similarly the latent heat flux isestimated by simultaneous measurement of turbulent velocity and humidity fluctuations.However, as Schmid et al. (1991) caution, the position of the instruments, both in thehorizontal and vertical, can significantly affect the meaning of flux measurements. Thesource area for the measured flux depends on the instrument height as well as on theturbulent characteristics of the flow.While the eddy correlation technique has been successfully used to measure the SEB66of rural and urban areas, its application is restricted in a park setting. To implement thistechnique the sensors must be positioned such that the source areas for the turbulentfluxes originate within the park. Therefore, depending on the size of the park, this mayrestrict the use of eddy correlation instrumentation to the centre of the park, if at all. Inaddition, again to meet source area requirements, the sensors have to be mounted closeto the surface. This may introduce under—sampling of the flux because the spectra andcospectra are truncated.Bowen ratio energy balanceThis method simplifies the evaluation of turbulent heat fluxes by taking the ratio of thesensible and latent heat fluxes and apportioning the available energy (Q4— QG) accordingto this non—dimensional ratio. The method relies on assumptions of constancy of fluxeswith height; a homogeneous surface and similarity of the transfer coefficients for heatand water vapour. The Bowen ratio (/3) is calculated from a profile of temperature andhumidity measurements:_L_cp 311QE L,,qwhere c, is the specific heat of air at constant pressure; 2T is the difference in meantemperature between measurement levels; and is the difference in average specifichumidity. From the energy balance, QH and QE are individually estimated:* *QE=GG (3.12)and sensible heat can be estimated by residual from:QH=Q*_QG_QE (3.13)The technique has been widely applied, particularly in estimating the SEB at rurallocations where fetch requirements are satisfied, and is used at the rural site in Vancouver67in the present study. In urban applications recent research suggests that this approachmay not be as suitable. Roth and Oke (1994) in a study of turbulence at a suburban sitein Vancouver, B.C., found that the transfer coefficients for heat and water vapour weredissimilar, particularly in slightly unstable and near neutral conditions. Further, thelimited fetch conditions prevailing in most parks again restrict the use of the approachgreatly.LysimetryAn alternative approach to measure QE is lysimetry. This technique is not hamperedby restrictions of several atmospheric methods. A lysimeter is a cylindrical containerin which an undisturbed block of soil and vegetation is isolated and its water budgetis carefully monitored. Changes in the mass of the lysimeter (gain due to precipitation, irrigation or dewfall; loss due to evapotranspiration) are monitored by weighing thelysimeter via a sensitive balance installed underneath or a manometer measuring differences of hydrostatic pressure. The dimensions of the lysimeter depend on the type ofvegetation, its depth, as well as on the observation time period. Large lysimeters with asurface area greater than 2 m2 are the standard instrument for measuring evapotranspiration (Slatyer and Mcllroy, 1961). However they are typically difficult to install, requiresophisticated engineering backup and consequently are very expensive.A recent development is the use of “mini—lysimeters” (diameter < 0.5 m) in fieldstudies (Clark et al., 1984; Dugas and Bland, 1989; Isard and Belding, 1989; Grimmondet al., 1992). As Grimmond and Isard (1992) discuss, mini—lysimeters have the followingadvantages over large lysimeters:• they permit measurement of QE from small areas;• they create less disturbance to the environment during installation;68• they are considerably cheaper to install and operate.Because of these characteristics, several mini—lysimeters can be used to study thespatial variability of QE. Hourly evaporation, B (kg m2 h’), is estimated from:B = p(F—4-) (3.14)where P is precipitation (m h1),1W is the change in weight (kg h’), A is the surfacearea (m2) and p is the density of water (kg m3) and the fluxes of water mass (B) andenergy (QE) are linked by:QE = LE (3.15)There are several potential sources of error in estimates of evapotranspiration fromthese instruments. For example, as the surface area decreases a smaller plant populationis sampled and the potential for differences in crop growth between the lysimeter and thefield increases. Secondly, errors can occur due to the “bloom effect”, whereby the areaof exposed plant canopy exceeds the lysimeter area. Thirdly, the non—vegetated edgebetween the soil—plant monolith and the surrounding area can increase turbulence in theair immediately above the lysimeter. Fourthly, the walls of the lysimeter may affect thethermal regime of the soil and vegetation. Finally, there may be errors associated withthe electronics and mechanics of the weighing device. In general these potential sourcesof error are inversely proportional to the surface area of the lysimeter (Dugas and Bland,1989).Grimmond and Isard (1992) compared the performance of mini—lysimeters with evaporative flux measurements using eddy correlation instrumentation over an extensive homogeneous surface. Their results show that QE, stomatal resistance and soil moisturemeasurements from the lysimeters compare favourably to independent measures takenoutside the lysimeters.69The mini—lysimeters developed by Grimmond and Isard (1992) are used in this studyto measure the spatial variability of evapotranspiration. Each lysimeter consists of a soilmonolith, enclosed in a container, that rests upon a load cell which continuously measuresits weight (Fig. 3.6). The specifications of the mini—lysimeter are given in Table 3.1.70Table 3.1: Specifications of the mini—lysimeters.Feature SpecificationWeighing mechanism Interface Inc single point I load cell model SP-I 50Internal diameter (m) 0.27Surface area (m2) 0.0573Depth (m) 0.25E resolution (mm h’) 0.0175QE resolution © 20°C (W m2) 11.6-_;- Ccxt&iec• ktemdConloinerWi’Figure 3.6: Design of the miniature lysimeter (after Grimmond and Isard, 1992).14 O.27mcell cle ad cI I71Chapter 4Surface Energy Balance of an Urban Park in a Hot Summer MediterraneanClimate— the Sacramento Study4.1 Observation programme4.1.1 Physical setting and observation sitesGeneral descriptionIn August 1991, a joint research programme involving research groups from the Universityof British Columbia, Indiana State University and MacQuarie University, Sydney, studiedthe surface energy balance of different land—use types in Sacramento, California.Simultaneous measurements of surface energy balance (SEB) fluxes and other climaticvariables were made at a suburban and two rural sites— one wet (irrigated sod farm)and one dry (scrub, grassland) (Fig. 2.1 and Fig. 4.1). A description of the observationprogramme is presented here; further details are in Crimmond et al. (1993). At the sametime, surface energy balance measurements were taken in a small irrigated suburban park.Continuous measurements of the SEB fluxes were taken from August 19 to 30, 1991 atthe suburban and rural sites and from August 22 to 28 at the park site. The average airtemperature for August was 22.9°C (compared with a normal 23.7°C), and it was drierthan normal with only 0.25 mm rain recorded in the six weeks prior to the measurementperiod, (compared with an average August precipitation of 1.8 mm).The synoptic conditions were otherwise typical for the region; cloudless skies prevailedby day with occasional night cloud. A cold front extended into the Sacramento region72CA)(a)Suburbanneighbourhood(b)OrvilleWrightPark(c)Wetruralsite(d)Dryrural siteFigure4J:PhotographsofeachstudysiteinSacramento.on August 25 giving some high cloud, and another front passed through on August 28.Site descriptionsPark measurements were conducted in Orville Wright Park, a small (3.6 ha) park in thesuburb of Mission Oaks. The park is mainly grass with a border of trees (5—10 m high).There is a large sandpit for volley ball and a children’s play area. The neighbourhood is aresidential area of predominantly single family dwellings with a well irrrigated (mesiscape)vegetation.The park is heavily irrigated three or four nights per week and the grass is keptcropped at 0.05—0.10 m. The park soils consist of a surface layer of brown, fine sandyloam to a depth of 0.15—0.30 m, below which is a clay hardpan. Consequently permeabilityis poor, and the park surface can remain damp well into the afternoon following a nightof irrigation.The suburban site was located in the residential neighbourhood of Carmichael, approximately 7 km northwest of Orville Wright Park. The land—use in this suburb issimilar to that surrounding the park. Irrigation of private gardens is permitted on alternate days.The wet rural site was at an irrigated sod farm about 18 km southwest of downtownSacramento in an area of intensive agriculture. Irrigation at this site occurred every 7—10days and the surrounding fields (for about 1 km x 1 km) were also regularly irrigated. Thegrass is maintained at 0.05 m. The dry rural site was about 18 km southeast of downtownSacramento with a surface cover of tall (0.5 m) unirrigated grass. The topography ateach site is generally fiat.Instruments were installed over several days:74• Suburban: August 19, year day, YD, 231• Dry rural: August 20, YD 232• Wet rural: August 21, YD 233• Park: August 22, YD 234.Continuous measurements were taken at the first three sites until August 29, with theexception of a six hour period on August 26 at the wet rural site, when irrigation wasin progress. Three sites were installed in Orville Wright Park on August 22, 1991 withmeasurements over the following seven days. Whilst there is a complete record for Site 3,Site 2 was later starting due to the need to relocate the lysimeter (a broken irrigation pipesaturated the first lysimeter pit). At Site 1 data were lost on August 26 (YD 236) andthe lysimeter failed on August 28 (YD 238). Due to these gaps, and given the essentialsimilarity of conditions over the period, two day averages (August 25 and 27) are derivedfor the purpose of analysis. While net radiation at the sites showed little variability fromthe average, the turbulent fluxes could range up to 30% from their average on an hourlybasis during the daytime.4.1.2 InstrumentationPark SiteThe surface energy balance at the park was sampled at three sites along a transect fromthe edge of the park to its centre. The sites are aligned in a southwesterly directionparallel to the direction of the climatological prevailing wind. Site 1 is at the edge of thepark 4 m in from the boundary; site 2 is 27 m from the edge and site.3 is near the centreof the park, 96 m from the boundary (Fig. 4.2).750NLSSSNSN________U)St.MarksWay_______aso000000000000—=Ooo0____C1?0oo%j100I0_H0o000o—0)V6o(1)VVct___:U)_NNNN.4GilaWayX_____________________________________________VNet all—wave radiation was measured at each site with a Swissteco (model Si) netpyrradiometer mounted 0.8 m above the ground. The ground heat flux was measuredat each site with two Middleton soil heat flux plates buried at 0.05 m depth. As no soiltemperatures were taken, the divergence of the heat flux between the surface and a depthof 0.05 m was estimated from theory’ (Leuning et al., 1982). The latent heat flux wasmeasured at the three sites by mini—lysimeters and the sensible heat flux estimated byresidual.Soil moisture is spatially variable in Orville Wright Park, partly due to the variabledepth of the hard pan, and also to proximity to the sprinkler heads. The lysimetermonoliths retained a similar amount of moisture to that of the surrounding soil. At theend of the observation period Site 1 had a soil moisture content in the lysimeter of 19%compared with 16% nearby; Site 2 had 17% both in and around the lysimeter; and Site3 had 24% compared with 20% nearby.Air temperatures were measured at all sites at a height of about 1.3 m with thermistorprobes, and at Site 3 a Vaisala capacitance sensor (model HMP 35C) also measuredrelative humidity. Wind direction at a height of 1 m was measured at Site 2 during thelatter half of the research period. Soil moisture was sampled along the transect everythree days. The fluxes and climatic variables were sampled every 20 seconds with 15minute averages recorded. Hourly averages are derived by averaging the four 15 minutevalues.In August 1993 a further two days of measurements were made in Orville WrightPark. Two sites were established — one in the centre of the park and the second over aconcrete driveway on a property west of the park. In the park Q* was measured with aSwissteco pyrradiometer (model Si), QH was measured by a CSI CA27 one—dimensional‘The ground heat flux at the surface is estimated from the ground heat flux at depth by consideringthe conduction of heat into soil which results from a sinusoidal temperature wave imposed at the surface.77sonic—anemometer—thermometer (SAT) system sampling at 10 Hz with covariances determined over 15 minutes. Wind speed and direction were measured with an R.M. YoungWind Sentry at a height of 1.5 m. At both sites air temperature and relative humiditywere measured with a Vaisala capacitance temperature and humidity probe and surfacetemperature was measured with an Everest 4000A infra—red thermometer. Net radiationand the climatic variables were sampled at 0.1 Hz with 15 minute averages recorded.Corrections to QH were made for density effects (Webb et al., 1980). Kristensen andFitzjarrald (1984) suggest that in unstable conditions it is not necessary to correct forline—averaging effects down to heights 4—5 times the length of the sonic path. The SAThas a path length of 0.1 m therefore allowing measurement down to 0.4 m. However, instable conditions QH may be underestimated.This set of measurements had the primary objective of verifying the earlier (1991)estimates of the sensible heat flux (by residual), as well as allowing assessment of thedriving forces for park evaporation (surface temperature difference and vapour pressuredifference between the park and the adjacent urban air).Suburban and rural sitesGrimmond et at. (1993) detail the instrumentation used in the observation programme.The following discussion is from their paper. Suburban flux measurements were takenat a height of 29 m. This is judged to be above the roughness layer and in the constantflux layer above the city (Oke et at., 1989). At the rural sites flux measurements wereat 1.3 m (wet) and at 1.8 m (dry). Both are considered representative of the microscale.At each site measurements of the turbulent fluxes were made using the eddy correlationapproach. Sensible heat was measured with a SAT and QE with a CSI KH2O Kryptonhygrometer. Air temperature, specific humidity and vertical wind velocity were sampledat 5 Hz with covariances determined over 15 minutes.78Net all—wave radiation was measured using a Swissteco miniature net pyrradiometerat the suburban site and Radiation Energy Balance System (REBS) net pyrradiometersat the two rural sites. Soil heat flux was measured at each site using two REBS soilheat flux plates with a CSI thermocouple system installed above the plates to accountfor thermal divergence between the plates and the surface. Heat storage at the suburbansite is estimated as a residual. Soil moisture was measured daily using a Soil MoistureEquipment Corporation Trase 6000 XI time domain reflectometry system.Slow response air temperature, relative humidity and wind speed and direction weremeasured at each site. These were sampled at 0.1 Hz and averaged over 15 minuteintervals. The energy balance fluxes and climatological variables are compared usinghourly fluxes which represent the average of the four 15 minute values obtained for eachhour.4.1.3 Probable errors in surface energy balance instrumentationThe instruments that were used to measure the same fluxes at different sites were calibrated and/or intercompared prior to installation and appropriate corrections incorporated into the results. The net pyrradiometers were calibrated by comparison with astandard by the Canadian National Radiation Laboratory (Atmospheric EnvironmentService). They should be accurate to within 3—4% with an even smaller inter—instrumentcomparison (Latimer, 1972). The mini—lysimeters were calibrated prior to use andchecked twice daily through the observation period. Measurements of QE from theseinstruments are precise to within 12 W m2 (Crimmond and Isard, 1992).It is difficult to assess the actual measurement error associated with eddy correlationestimates of QH (Cleugh, 1988). An intercomparison of hourly fluxes from two instruments found a RMSE of 12 W m2 (Cleugh, 1988; Schmid, 1988). Grimmond (1988)79reports that the measurement error is probably less than 10%. The Krypton hygrometers were calibrated by comparison of vapour density measurements with a dew—pointgenerator. Flux corrections were made for oxygen absorption by the hygrometer and forair density (Webb et al., 1980; Tanner and Greene, 1989); and at the suburban site forfrequency response and spatial separation of the eddy correlation sensors (S. Grimmond,pers. comm., 1993, and M. Roth, pers. comm., 1992).Both the sensible heat flux at the park sites and the storage heat flux in the suburbanSEB are estimated as residuals. This has the inherent problem of accumulating themeasurement errors of the other energy balance components.4.2 Urban park surface energy balance (SEB)The August 1993 measurements of surface and air temperatures and humidity both inthe park, and over a driveway adjacent to the park, serve to illustrate the surface climatedifferences. (Figure 4.3).Air temperature shows little difference between the two sites although the park isconsistently cooler during the daytime, but only by 1—1.5°C. The park is always morehumid particularly after sunrise when dew is evaporating. The park surface is alwayscooler than the adjacent paved surface with a maximum difference of 16.6°C occurringat 1500 LAT. This strong contrast in surface temperatures could establish warm airadvection over the cooler park.4.2.1 Energy balance partitioningFigure 4.4 shows the average (2—day) diurnal surface energy balance at each of the parksites. The energy balance at each is dominated by QE, and Sites 2 and 3 show similarpartitioning of energy. Net radiation peaks at about 530 W m2 and QE is very high80a) Temperature and relative humidity differences b) Vapour pressure difference0 500 1 000 1 500LAT20 011RH—01-—.-- T / ,—.. —0.3- --C10 /—0.5-o‘—(1) 5N/_____________- -0.7o —— I —0.9-o0 -—1_i‘ —5 4 •-4 •-,0a > —1.3-- .•• I”t -—10—1.52000LATFigure 4.3: Differences in diurnal air and surface temperature, T and T (°C), relativehumidity, RH (%) in (a), and vapour pressure, LWP, in (b), between adjacent park andpaved sites at Orville Wright Park, Sacramento.with a maximum of 400—500 W m2 in the early afternoon. The sensible heat fluxpeaks near 150 W m2 just before midday, and in the afternoon becomes negative (i.e.directed towards the surface). The ground heat flux is very low; almost negligible. TheSEB at Site 1 clearly shows the influence of edge effects. Firstly, the shadows fromthe nearby trees and hedge create a dip in Q* at 1300—1500 LAT. Secondly, QE is veryhigh — up to 600 W m2 and frequently in excess of Q* Consequently QH is oftennegative, supplementing Q* in achieving these high evapotranspiration rates, especiallyin the afternoon. A statistical summary of the daily and daytime (Q* > 0) fluxes ispresented in Table 4.1.The net all—wave radiant flux density is conservative across the park; each site hasa daytime average of approximately 330 W m2. At Site 1, Q* dips in the early afternoon due to shading. The latent heat flux shows both spatial and temporal differenceswith increasing fetch over the park. QE is greatest at Site 1 due to microscale advection81Figure 4.4: Two—day average surface energy balances at three park sites in Orville WrightPark, Sacramento in August, 1991. The net radiation and sensible heat fluxes measuredat Site 3 in August 1993 are also shown for comparison. For the 1991 measurements, netradiation was within 10% of the average on an hourly basis; QE within 20% and QH andQG within 30%. In 1993, Q* varied less than 4% and QH was within 25% of the averageon an hourly basis.a) Site 1 b) Site 2\ 4Q.V‘V\7.-- -v—v -c- _x-.< \.,700600500E 400300>.6 200ci,0 100xD= 0!—G-—Q) : :•1 ,÷.+•_V. - + - ÷—x— V —700600500400300>s&5 200ci,t’ 1000i—300700600500400300- 200Cci)100xZZ 0Izi_ -.500 1000 1500 2000LATd) Site .3 (1 993).,. -p..4 ,. .-. : -•,.‘I .4500 1 000 1 500 2000tATc)Site3—4-— 0•v OH3_v._w-._..Al Oc.4-+.\—30007006005004003002001000—100—200E>‘cciCIi,D>‘crcci). I)I.+ +.0 500 1000 1500 2000LAT0 500 1000 1500 2000LAT82Table 4.1: Daily and daytime (Q* > 0) averages and ratios for energy balance fluxes atsites in Orville Wright Park. All fluxes are in W m2.Site Q* QE QH QG = a =E Eqa) DAILY1 130 196 —61 —5 1.51 —0.47 —0.31 1.992 114 147 —29 —3 1.28 —0.25 —0.20 1.653 119 142 -16 —7 1.19 —0.13’ —0.11 1.52b) DAYTIME1 327 374 -64 16 1.15 -0.20 -0.17 1.672 333 302 16 15 0.91 0.05 0.05 1.303 335 302 26 7 0.90 0.082 0.09 1.27across the moist park boundary, supplying an additional energy source to drive evapotranspiration. The peak occurs at 1500 LAT which corresponds to the time of maximumforcing due to surface temperature differentials between the park and the adjacent pavedsurfaces (see Figure 4.3). Further towards the centre of the park, the magnitude of QEdecreases by about 20% (Table 4.1).An appreciation of the advective influence can be gained by comparing the measuredlatent heat flux, QE, to the latent heat flux which could occur in the absence of advection,Q.q, and Priestley and Taylor’s (1972) potential evaporation, QE (if a > 1.26 thenQE> QE) (Table 4.1 and Fig. 4.5). Except for a few hours in the morning, evaporationat all sites easily exceeds the equilibrium rate, indicating the influence of mesoscaleadvection. As QE > QEp at all sites in the afternoon, there is evidence of enhancedadvective effects. Since evaporation continues through the night daily a is even higher83than the daytime ratios. On a daily basis evaporation at the edge of the park is doublethat of Qq. This decreases to 1.65 at Site 2 and 1.52 in the centre of the park. Thesimilarity of Sites 2 and 3 suggests that the major influence of fetch is confined to a stripwithin 25 m of the park edge. The influence of fetch and the contribution of the edgeand oasis effects to this advection are considered further in subsection 4.2.2.In the 1991 park study QH was estimated as a residual while in the 1993 observationperiod it was directly measured by the eddy correlation technique. The diurnal trend andthe resulting daytime average of both QH and the ratio QH/Q* in 1993 are very similarto those observed in 1991 (see Fig. 4.4 and Table 4.1) and confirm that estimating QHby residual is a satisfactory procedure. Sensible heat is positive (into the air) in themorning but as the surface temperature differential increases between the cool park andthe warmer neighbourhood in the mid—afternoon, warm air advects over the park andQH becomes negative, (directed towards the surface).The fluxes normalized by Q* reveal daytime trends in energy partitioning by the park(Fig. 4.6). At Sites 2 and 3, QE/Q* shows a steady increase throughout the daytime. Theincrease of this ratio suggests that evaporation is not water restricted but is limited byenergy availability. Thus it increases in the morning with the net radiation and continuesto increase as sensible heat becomes an additional energy source in the late afternoon.The corresponding trend of QH/Q* decreases throughout the day and becomes negativein the afternoon.The Bowen ratio (/3) for 0800—1600 LAT shows a general decrease at all sites. Thelow values reflect the dominance of the latent heat flux with QE using two to three timesas much energy as QH. At Site 1 the daytime average of —0.17 confirms that there is astrong edge effect (Table 4.1). The time of reversal in /3 becomes later further into thepark (negative at Site 2 by 1400 LAT, and 1600 LAT at Site 3). On a daytime basis theBowen ratio at these sites is as small as at an open water site (< 0.10).84a) Site 1 b) Site 2700 700600 600500 500E E400 400>‘300C0 300C-ox 200 ° 200x=100 100i5Cij Q Cw 0—100—100c) Site 3700600soE300C0)0400 /•+x 2002100aC___________________________________________U 0—1000 500 1000 1500 2000LATFigure 4.5: Comparison of the measured QE with and QEp at three sites in OrvilleWright Park, Sacramento in August, 1991.85[AT [AT0zo -1Figure 4.6: Two—day average of the SEB flux ratios for daytime hours at three park sites.To facilitate interpretation, the data for Site 1 near midday are omitted due to shadingby nearby trees.a) Qe/Q* b) OH/Q320--4-- 2--,<-- 3j -.- -- -:_X.—Z-—---:‘“.:--1--4-- 2-—x-- 30‘1)00r0—2800 1000 1200 1400 1600 1800LATc) owen rafio800 1000 1200 1400 1600 1800LAT1 5800 1000 1200 1400 1600 1800LAT864.2.2 The influence of fetch on QEA strong edge effect is observed at the park with very high rates of evaporation near theupwind boundary. The earlier discussion of the spatial variability of QE (Section 4.2.1)did not incorporate the true fetch because fixed sites aligned along a transect are notnecessarily parallel to the wind direction. Figure 4.7 is a plot of the daytime average QEagainst true fetch for August 27 (YD 239); when wind direction data are available. Itshows a decrease in QE from the edge of the park to its centre (Fig. 4.7). The similarityof QE at Sites 2 and 3 suggests that the edge effect is confined to about 20 m from theupwind edge. At Sites 2 and 3 there is no evidence for an oasis effect as QE < Q butevaporation is above the equilibrium rate, indicating some mesoscale advective influence.While the park does not exhibit a strong oasis effect when averaged on a daytimebasis, both the daily average c and the diurnal variation of QH indicate that the parkexperiences both edge and oasis effects (Table 4.1 and Fig. 4.5). From 1600 until atleast 2200 LAT, QE exceeds QEp by up to 330 W m2 at Site 1, 80 W m2 at Site 2,and 100 W m2 at Site 3. Thus during these hours a strong edge effect (130 W m2) issuperimposed upon a considerable oasis effect (about 90 W m2).The evapotranspiration trends in Orville Wright Park agree well with the conceptualmodel of an exponential decay of QE with distance from a leading edge in a well—irrigatedpark (Fig. 3.1). However, in a park of this size, the oasis effect is only present in theafternoon and evening. There is little evidence of the turbulence regime of the parkaffecting the decay of QE. Perhaps Sites 1 and 2 are in a quiet zone, sheltered by thesurrounding vegetation and dwellings. There is some evidence to support this contention.Air temperatures are slightly higher at these sites,Q* is slightly reduced and QG is slightlyincreased. However, evapotranspiration is not reduced at these sites although withoutthe shelter, the eddy diffusivity near the park edge would increase and QE could be874004.2.3 Summary of the energetics of a suburban park in a hot summer Mediterranean climate.In the well—irrigated Orville Wright Park set in a hot climate, the surface energy balanceis dominated by evaporation which acts as the main sink for net radiation. The groundheat flux is negligible. Very high rates of QE are measured in the park particularly at itsupwind edge. The decay of QE with increasing fetch generally fits the hypothetical edgeeffect model, except the vegetated park border may disturb the flow and create a quietzone in its lee. Sensible heat is a moderate sink for Q* in the morning however, by earlyafternoon QH turns negative, establishing high rates of QE and allowing the park to actas an “oasis”.A microscale advectionedge effect__.‘- mesoscale advection380360340320300280260240220200__QE___0 10 20 30 40 50 60Fetch (m)Figure 4.7: The variation of average daytime QE with distance of fetch from the parkedge for August 27 (YD 239). The average wind speed for this period was 2.84 m s1 atthe nearby suburban site.considerably enhanced.884.3 The urban park in its context — park, suburban and rural comparisons4.3.1 Energy partitioning in different land—use typesThe 2—day average energy balances for each of the four land—use types are given inFigure 4.8 and Table 4.2. The park data refer to Site 3 — the centre of the park. Thefollowing analysis compares the energy balance of the park to the others (for comparisonsbetween the suburban and rural sites see Grimmond and Oke, 1993).At the rural sites it is theoretically possible to achieve energy balance closure, becauseeach flux is measured directly. However, because of measurement errors associated witheach flux, few studies ever obtain true closure. Further corrections could be made (e.g.for the frequency response and the spatial separation of eddy correlation sensors) butsince the unaccounted residuals are relatively small (5% of Q* at the wet site and 1% atthe dry site on a daily basis) no corrections were made.In energetic terms the park is most like the wet rural site (Fig. 4.8 and Table 4.2).Both receive a similar net radiation input and have an energy balance dominated byevaporation; although QE is much higher (140%) in the park. The sensible heat fluxfor both sites peaks around 100 W m2 and turns negative in the late afternoon. Atboth sites QG is negligible. In contrast, the SEB of the suburban site (whose spatialaverage must include parks) is dominated by QH which peaks near 300 W m2, whereasQE is much lower at 150 W m2. The suburban storage flux is positive by day reaching170 W m2 in the mid afternoon. The energy balance at the dry rural site is almostcompletely dominated by QH which peaks around 400 W m2, QE is negligible and Qcplays a more important role, peaking around 80 W m2.The 2—day averages of the turbulent fluxes (Table 4.2) show distinct differences inenergy partitioning between sites. Daily evaporation in the park is 3.16 times higherthan from the nearby suburban neighbourhood. Few studies have reported comparisons89Table 4.2: Daily and daytime 2—day averages and ratios of energy balance fluxes fordifferent land—use types in Sacramento, CA.Site Q* QE QH Qa LQs ResWm2 ‘Wm2 Vm2 Wm2 Wm2 ‘Wm2a) DAILYPark 119 142 -16 -7 1.19 —0.13 —0.11 1.52Wet rural 128 108 17 -3 6 0.84 0.13 0.16 1.25Suburban 107 45 73 -11 0.42 0.68 1.62 0.53Dry rural 110 1 105 2 2 0.01 0.95 78 0.02Park/Suburban 1.11 3.16 -0.22 0.64Park/Wet rural 0.93 1.31 -0.94 2.33b) DAYTIMEPark 335 302 26 7 0.90 0.08 0.09 1.27Wet rural 342 214 52 20 55 0.63 0.15 0.24 1.03Suburban 304 89 165 50 0.29 0.54 1.85 0.49Dry rural 307 4 253 34 16 0.01 0.82 57 0.02Park/Suburban 1.10 3.39 0.16 0.14Park/Wet rural 0.98 1.41 0.50 0.3590Figure 4.8: Average energy balances for different land—use types in Sacramento, CA.,based on two days observations in August, 1991. Net radiation is within 10% of themean on an hourly basis; QE within 20%; and QH and QG within 30%.91o) Suburban600b) ParkV\‘\ --1-- QuF! -‘: ‘.+-+ + -<-- )-.‘->-ox>‘CwE>‘(I)C0x>s0CLJ500 500::200 J- ‘\ 200100-1000+ j- 0—100 Ui —100—2000 500 1 000 1 500 2000LATc) We rural600500—e— Q400._..VV— Qc E300 7200100 y__- .--x- -----A-’—2000 500 1000 1500 2000IATd) Dry rural600 I500400:::10:—100—100—200 -0 500 1 000 1 500 2000LAT—200500 1 000 1500LAT2000654—‘ 3EE0loLATFigure 4.9: Cumulative evaporation for the different land—use types in Sacramento, CA.Data are two—day averages.of evaporation over moist greenspace compared to the suburban neighbourhood. Oke(1979) in a study of advectively—assisted evapotranspiration from an irrigated lawn inVancouver, B.C., found evaporation from the lawn was 3.09 times of that in a similarsuburban neighbourhood. The differences in the evaporation rates in Sacramento arehighlighted in Figure 4.9, which shows the cumulative evaporation for each site. Thepark has a daily evaporation rate of about 5 mm closely followed by the wet rural site at4.2 mm. Evaporation from the suburban site is much lower at 1.6 mm and it is negligibleat the dry rural site. The dominance of QE at the park and wet site is probably a reflectionof similar moisture regimes. Both have high soil moisture due to heavy irrigation andare influenced by advective enhancement of evaporation.The differences in energy partitioning between the sites are clearly seen in an examination of the fluxes when normalized by Q* (Fig. 4.10 and Table 4.2). The daytimeratio of QE/Q* again shows the dominance of QE at the park (90%), followed by the wet500 1 000 1 500 200092rural site (63%), suburban (32%) and dry rural (1%). The proportion of radiant energydirected into latent heat remains essentially constant at each site until the mid— to lateafternooon. Then there is an increase in the ratio at the park and wet sites, corresponding to a decrease in the ratio Qff/Q*. At these sites the sensible heat flux changes signin the late afternoon indicating advectively—assisted evaporation that elevates QE aboveQ* This advective component is most important at the park site where QH is negativeby 1500 LAT and QE exceeds Q* for the remainder of the afternoon and evening. Theadvection at this site is sustained by the supply of warm urban air resulting in the parkacting as an “oasis”.The Bowen ratio values (/3) illustrate the interplay between the turbulent sensibleand latent heat fluxes at each site. The park and wet rural sites have similar decreasingtrends in this ratio throughout the day (Fig. 4.10). The daily averages for /3 are —0.11for the park and 0.16 for the wet rural site (Table 4.2). This value for /3 for the park isclose to that observed by Oke (1979) for an irrigated suburban lawn in Vancouver, B.C.(-0.13).At the suburban site where QH is more dominant in the energy balance, /3 is morevariable and has a daily average of 1.62. For a longer averaging period (10 days) at thissite Grimmond and Oke (1994) observed a /3 of 1.20. This is comparable to Bowen ratiosreported for other cities located in semi—arid environments (eg. 1.33 in Los Angeles and1.35 in Tucson, (Grimmond and Oke, 1994)).The ratio QG/Q* is almost negligible at the park and wet rural sites (2% and 6%respectively) although at the dry rural site QG is 11% of Q*. The storage heat flux atthe suburban site has a daytime average of 16% of Q*• Throughout the day this fluxincreases in importance and in the afternoon accounts for about 30% of Q*•93a)QE/Q b)QH/Q1.5 •!.lI.I.I,-—I’1• ‘II1.0Dry— —— — — — — — —- Suburban0.5 . 4--__4.0.0/v.t—0.5—1.1700 900 1100 1300 1500[ATc) Bowen ratio34+/ \,/. / \2 +. / 4s Suburban---+. •1\4P700 900 1100130015001700[ATFigure 4.10: Daytime flux ratios for different land—use types in Sacramento, CA. Dataare two—day averages. The Bowen ratio for the dry rural site is not shown as it is about60.94[ATPark4.3.2 Summary of the urban park surface energy balance compared to otherland—use types in a hot summer Mediterranean climateThe irrigated urban park has a surface energy balance most similar to a wet rural site.Both have energy balances dominated by evaporation and consistently negative sensibleheat advection in the late afternoon. Soil heat flux at these sites is small. The dominanceof QE is a direct reflection of the similar soil moisture regimes due to irrigation. However,the park is distinguished from other land—use types by its propensity to act as an “oasis”.In a hot dry climate urban parks can have evapotranspiration rates in excess of thoseat irrigated rural sites. The Sacramento park had a daytime QE 1.41 times higher thanthe irrigated rural site and 3.39 times higher than the suburban value. This elevatedevapotranspiration is probably due to oasis—type advection of warm dry urban air overthe park supplying additional energy for evapotranspiration. In contrast, the surfaceenergy balance at the suburban and especially the dry rural site, is dominated by QH.95Chapter 5Surface Energy Balance of an Urban Park in a Cool Summer MediterraneanClimate— the Vancouver Study5.1 Observation programme5.1.1 Physical setting and observation sitesGeneral descriptionIn July 1992, simultaneous measurements of the surface energy balance were made overthree different land—uses in the Vancouver region. Measurements of the surface energybalance and other climatic variables were made at a park, a suburban and a rural site(Fig. 2.2 and Fig. 5.1). The rural and urban sites were installed in mid July and thepark sites on July 20. From July 20 to July 24 a slow—moving low pressure system waslocated off the coast of British Columbia. A cold front associated with this system passedover Vancouver on July 21 bringing cloudy skies and showers. The showers continued onJuly 22 with 11 mm of precipitation. Due to this unsettled weather the rural and urbansites were not brought on—line until July 25. By July 27 a high pressure system becameestablished giving mainly sunny skies for the remainder of the study period.Site descriptionsThe park selected for study was Trafalgar Park in the Arbutus district of Vancouver. It isa 4.86 ha open grasssed park with a partial border of mixed deciduous and coniferous trees(about 3—8 m high). Immediately to the west is Trafalgar Elementary School which has96CD(a)“Sunset”suburbanarea(b)TrafalgarPark(c)“Delta”ruralsite.LFigure5.1.:PhotographsofstudysitesinVancouver.a gravel playing field bordering the park. The park consists of several playing fields andhas a childrens playground and a fleidhouse. It is located in a residential neighbourhoodof predominantly one— and two—storey dwellings. Two blocks to the west is a smallcommercial strip. The neighbourhood is in a slight topographic depression.The park was irrigated until 3 weeks prior to the measurement period. At this time aban on sprinkling of lawns was implemented, so irrigation of the park ceased. However,11 mm of precipitation fell on July 22, two days after installing instrumentation in thepark. The grass was maintained at approximately 0.05 m height. The park was mownhalfway through the observation period.The suburban site was at “Sunset” in South Vancouver at a location used previouslyfor several energy balance studies (eg. Roth, 1991; Grimmond, 1988; Cleugh, 1988). Theneighbourhood consists of predominantly one— and two—storey dwellings. Throughoutthe measurement period there was a complete ban on irrigation of lawns. Irrigation ofgardens was permitted on alternate days.The rural site was in a mixed agricultural area of Delta, 20 km south of Vancouver.Measurements were conducted over a bush bean crop about 0.5 m high. The upwindfetch for the prevailing southwesterly wind was over beans for approximately 300 m andthen over mixed crops. There was no irrigation during the observation period but the sitereceived 7 mm of precipitation on July 22. Bean harvesting began earlier than anticipatedon July 29 and was completed by July 31. Measurements of climate variables were madeat a nearby tower (3 m high) at the junction of fields containing wheat and corn. Thesurrounding topography is very fiat delta land.985.1.2 InstrumentationPark siteIn Trafalgar Park, five sites were established in a transect from the southwest side tothe centre of the park, aligned in the direction of the prevailing wind. The sites werenumbered 1 to 5; 1 being closest to the edge of the park and 5 in the centre (Fig. 5.2).At Sites 1 and 5, net all—wave radiation, Q*, was measured with a Swissteco netpyrradiometer (model Si) mounted at 1 m and soil heat flux, Q, with Middleton soilheat flux plates buried at a depth of 0.05 m. Thermocouples were installed above eachplate to account for thermal divergence between the surface and the depth of the plates.The latent heat flux, QE, was measured at all sites using mini—lysimeters as described inSection 3.3.The sensible heat flux, QH, was measured at Site 5 only, by a CSI CA27 SAT samplingat a rate of 10 Hz. This instrument, mounted at a height of 1 m, gives an integratedsensible heat flux for the fetch over the park rather than the heat flux of that point.When the prevailing wind is from the southwest there is a fetch of 100 m over the park.Estimates of QH at Sites 1—4 were found by residual in the surface energy balance (ie. apoint measurement given by (Q*— (QE + QG)).At Site 1 air temperature was measured at a height of 1.5 m using a thermistor probemounted in a radiation shield. Surface temperature was measured using an Everest4000A infra—red radiation thermometer, precipitation by a tipping bucket raingauge andat Site 4 surface wetness by a Weiss surface wetness sensor (Weiss and Lukens, 1981). AtSite 5 air temperature and relative humidity were measured using a Vaisala capacitancetemperature—humidity probe mounted in a radiation shield, at a height of 1.5 m. Windspeed and direction were measured at a height of 2 m by an R.M. Young 03001 cupanemometer and wind vane set. Soil moisture was sampled along the transect every9923rd Ave.24th Ave.Alamein AveC,,1Oliver Cr.Figure 5.2: Location of sites in Trafalgar Park, Vancouver.100three days for gravimetric analysis.With the exception of the SAT, all variables were sampled at 0.1 Hz with 15 minunteaverages recorded by CSI CR21X data loggers. Hourly averages of the energy balancefluxes and climatic variables are derived from an average of the four 15 minute valuesobtained for each hour.The mini—lysimeters were installed on July 21 (YD 202) with continuous observationsuntil July 31. Some difficulty was encountered during installation. The high sand content of the park soils resulted in some fragmentation of the lysimeter monoliths causingdisturbance. By the end of the observation period soil moisture was depleted in thelysimeters compared to the surroundings. Site 1 had dried out the most (with 11% soilmoisture in the lysimeter after 10 days compared to 27% nearby) while Site 5 had theleast depletion (20% in the lysimeter compared to 24% nearby).Since the latent heat flux was both measured (lysimeter) and estimated (SEB residualusing QE = Q*_QG_QHEC) at Site 5, these data were used to determine if any reductionin evapotranspiration had occurred due to moisture stress by the grass. Evaporationfrom the mini—lysimeter declined throughout the observation period, whereas that by theresidual increased (Fig. 5.3). Therefore it appears the monolith became unrepresentativeexcept for the first few days when the two estimates agree well. This comparison suggeststhe residual method should be used for Site 5.To see if the other lysimeters were similarly affected, time trends of the averagedaytime QE at each site are compared (Fig. 5.4). This reveals that sites 1—3 remainapproximately in the same relation to the areal (SEB) estimates until after YD 207. Butthe lysimeters at Sites 4 and 5 show signs of drying out. Therefore data from Site 4 areomitted from analysis and evaporation at Site 5 is estimated by SEB residual, ratherthan by the lysimeter. Further analysis of elements considers 3—day averages only for theperiod YD 205—207. Over these days the park is in a drying phase following 11 mm rain101400300200100—100—200Figure 5.3: A comparison of QE as measured by a mini—lysimeter and estimated as theSEB residual for Site 5 in Trafalgar Park.rain on YD 203 so there may be considerable variability in the fluxes about their mean.Therefore the results are limited in their generality.Suburban siteAt the suburban site instruments were mounted at two levels on a 33.5 m tower. Atthe top level (at an effective height of 19 m above zero—plane displacement) the sensibleand latent heat fluxes were measured using the same eddy correlation equipment as thesuburban Site in Sacramento (i.e. a SAT and a Krypton hygrometer). Signals weresampled at 10 Hz and the covariances determined over 15 minute intervals. Correctionswere applied to the resultant fluxes to account for oxygen absorption by the sensor, airdensity effects and for the frequency response and spatial separation of the sensors. Netall—wave radiation was measured using a Swissteco net pyrradiometer mounted at the204 205 206 207 208 209 210 211 212 213 214YD1021.3 I1.20.3 -.—0.2 I I I I I I205 206 207 208 209 210 211 212 213YDFigure 5.4: Time trend of daytime average QE/QEres for Sites 1-5 in Trafalgar Park,Vancouver.same height.At the lower level (1]. m effective height) air temperature and relative humidity weremeasured with a Vaisala capacitance temperature and humidity probe and wind speedand direction with a Met—One system comprising a wind vane (Met—One 024A) and cupanemometer (Met—One 012A). Precipitation was monitored with a tipping bucket raingauge and surface wetness with a Weiss wetness sensor placed near the raingauge, on theroof of the trailer, at the base of the tower.The climate variables were sampled at 0.1 Hz with 15 minute averages recorded on aCSI CR21X data logger. Hourly averages were derived for the energy balance fluxes andthe climate variables. Anthropogenic heat was neglected and L.Qs was estimated as theresidual in the energy balance (ie. Q*— (QE + QH)).103Rural siteA CSI Bowen Ratio system was used at Delta to measure the turbulent heat fluxes overthe bean crop. Humidity was measured with a cooled mirror dew point hygrometer(Model Dew—lO General Eastern Corp.). Air samples drawn from heights of 1 and 2 mwere routed to the cooled—mirror after passing through mixing volumes. The resolutionof the dew point temperature measurement is ±0.003° C over a ±35°C range. Given thestability of the Dew—lO, approximately 0.05°C, this yields a resolution of better than±0.001 kPa in vapour pressure. Every two minutes the air drawn across the cooled—mirror is switched from one measurement level to the other. The mirror is given 40 s tostabilize to the new dew point and an 80 s sample for each level is obtained during eachtwo minute cycle. Dew point temperature is measured every second and vapour pressureis calculated using the equation given by Lowe (1976). Air temperature is measured at1 and 2 m with chromel—constantan thermocouples. The differential voltage is due tothe difference in temperature between the temperatures at heights 1 and 2 m and has noinherent sensor offset error.This Bowen ratio (BR) system has been compared to others and to eddy correlationsystems in the field (Dugas et al., 1991). Four BR systems were compared: one CSI BRsystem as described above; two using reversing arms with dry— and wet—bulbs; and onewith dry— and wet—bulbs on fixed arms. The Bowen ratio (/3) values and calculated QEfrom the 4 BR systems were in close agreement. Differences of /3 were approximately±0.1 in the morning and afternoon and ±0.02 around noon and within 10% for the twodays. The eddy correlation value of QE was significantly (20—30%), and consistently, lessthan that from the BR system.Net all—wave radiation was measured with a Swissteco net pyrradiometer at a heightof 2.5 m. The soil heat flux was measured with two Middleton heat flux plates buried104at a depth of 0.05 m; one under a bean bush, the other between rows of beans. Bothwere corrected for vertical heat flux divergence effects. Surface temperature was measuredwith an Everest 4000A infra—red thermometer, mounted at 2.5 m. Surface wetness sensorswere placed both on the crop and on the bare ground, between the beans. Backgroundclimate variables measured at a nearby site (approximately 300 m from the Bowen ratiotower) included air temperature and relative humidity using a Vaisala temperature—humidity probe; wind speed and direction with an R.M. Young 03001-10 wind sentry set;and precipitation using a tipping—bucket raingauge. The sampling rate of these climatevariables was 0.1 Hz with 15 minute averages recorded on a data logger. Hourly averagesfor the energy balance fluxes and climatic variables are later derived.All instruments were calibrated and/or intercompared before use in the field. For anestimate of errors associated with these instruments see Section Urban park surface energy balance (SEB)5.2.1 Background climateAir and surface temperatures increased throughout the observation period as the weathercleared. The 3—day average vapour pressure differences between near—surface and overlying air in the park, together with park—surround surface temperature differences forYD 207, are plotted in Figure 5.5. The surface temperature difference between the parkand the adjacent gravel field is available only at three hour intervals. The surface temperature difference which is an important driving gradient for evaporation and advection,peaks in the early afternoon. The vapour pressure difference is generally small and thenear—surface park air has a slightly higher vapour pressure from 1000—1600 LAT.The wind regime in the park is influenced by both synoptic— and local scale systems.105a) Surface temperature difference b) Vapour pressure difference1211- 1.3a10 09 0.9Z5’ 6 e 0.77o 0.5a. 00.3> 0,o 0,4 0.1a.f- 3 0.1.3 02 0.-0.3_________________________________________________—0.5500 1000 1500 2000LATFigure 5.5: Surface forcing for evaporation in Trafalagar Park, Vancouver. The surfacetemperature difference between the park and the adjacent gravel field is shown in (a),with the vapour pressure difference between the near—surface park air and the overlyingair shown in (b).When skies cleared after the frontal disturbance, a land/sea—breeze system became established giving southwesterly winds in the afternoon, with a peak speed near 2 m s’.At night, very light east or southeasterly breezes were typical.5.2.2 Energy balance partitioningThe three—day average SEB for Sites 1-3 and 5 in Trafalgar Park are given in Figure 5.6.Table 5.1 is a statistical summary of the daily and daytime fluxes. Due to cloud, thenet radiation is less than the maximum possible and asymmetric about solar noon. Itvaries little across the park, with a daytime average of about 222 W m2. The latentheat flux is clearly dominant at all sites peaking near 220—270 W m2 and exceedsQ*for a brief period in the late afternoon. Sites 1, 2 and 5 have similar energy partitioningwith 67—71% of Q* directed into QE; about 20% going into QH and 8—13% into QG. The8AALAT106ground heat flux is more important at Site 1, at the expense of QE. Site 3 is characterizedby higher QE that accounts for 80% ofQ*• Consequently at this Site QH is the lowestof the set.Table 5.1: Daily and daytime (Q* > 0) energy balance fluxes and and ratios for sites inTrafalgar Park. All fluxes are 3—day averages expressed in W m2.Site Q* QE QH QG = =a) DAILY1 110 78 28 4 0.71 0.25 0.03 0.36 1.072 82 25 3 0.74 0.23 0.03 0.31 1.123 95 14 3 0.85 0.12 0.02 0.14 1.275 111 84 25 3 0.75 0.22 0.02 0.29 1.13b) DAYTIME1 221 149 44 29 0.67 0.20 0.13 0.30 1.132 155 45 23 0.70 0.20 0.10 0.29 1.143 178 26 19 0.80 0.12 0.08 0.15 1.295 223 158 46 19 0.71 0.21 0.08 0.29 1.13Although spatially variable, the turbulent fluxes show a similar temporal trend. Atall sites QE increases rapidly in the morning (probably due to evaporation of dewfall)and peaks around 1500 LAT coincident with the time of maximum net radiation. In thelate afternoon QE exceeds Q* for a few hours. In the morning the sensible heat flux isinitiaily low and possibly negative as dewfall is evaporated, but thereafter increases andremains at 70—100 W m2 for several hours in the early afternoon. It then decreases withQ* and may turn negative for a few hours in the late afternoon, enabling evaporation toexceed net radiation. This effect is the most pronounced at Site 3.The cumulative evapotranspiration (E) has a maximum at Site 3 (Fig. 5.7) wheredaily average evapotranspiration is close to 3.5 mm. At Sites 2 and 5 cumulative E is107Figure 5.6: Surface energy balance for Sites 1-3 and 5 in Trafalgar Park, Vancouver.Data are three—day averages. There is considerable variability of the fluxes about theirmean because the weather was clearing. On an hourly basis the fluxes may vary by upto 30—40% of their mean value.108a) Site 1 b) Site 24003002001 0003:>‘-o>‘G)wE>‘0C0)0x>‘CLiLAJ400300E, 200>0C,-0 100)<>‘0i5 0CLi—100 —100c)Site3 d)Site5400 I 400:200 200 / / \100 / 1000—100 —1000 500 1000 1500 2000 0 500 1000 1500 2000LATLAT LAT432EEU 10Figure 5.7: Average daily cumulative evapotranspiration for Sites 1—3 and 5 in TrafalgarPark, Vancouver.3 mm, while Site 1 is about 2.8 mm. This trend of an increase in QE through Sites 2and 3 and then a decrease to Site 5 reflects, in part, the turbulent regime of the park(Section 5.2.3).Analysis of the average daytime flux ratios reveals some diurnal trends (Fig. 5.8).Initially the ratio QE/Q* is high possibly due to the evaporation of dewfall in the earlymorning. This process would be assisted by the increase in both net radiation and windspeed soon after sunrise. At this time QE may exceed Q* but for the remainder of theday QE/Q* is fairly constant. Due to this high ratio in the morning, QH/Q* is initiallylow (and possibly negative) but becomes more constant around 0.2 through the day.On most of the mornings dew was freely available, but by noon this moisture is exhausted and transpiration rather than evaporation drives the moisture flux. Evaporationat all sites exceeds the equilibrium rate, indicating some mesoscale advective influence.However, since QE < QEp in the centre of the park, there is no evidence of enhanced0 500 1000 1500 2000LAT1090.—0.1—0.2-0.3—0.4 -500a) QE/0 b) QH/Q001// - 1-c-. 3-4— 50000700 900 1100 1300 1500 1700LATc) Bowen ratio700 900 1100LAT‘C•x- 2--q-- 3-4— 51300 1500 1700—0.1—0.2—0.3—0.4 —500 700 900 1100 1300 1500 1700LATFigure 5.8: Surface energy balance flux ratios at sites in Trafalgar Park. Data arethree—day averages.110230220 -210200 -190 •.. Q, — —180 . . -1 70 mesoscale . -A advection160A150140 -130 -120 I I I I0 20 40 60 80 100 120 140 160Fetch (m)Figure 5.9: The variation of QE with distance of fetch from the park edge based onthe daytime average for July 24-25, 1992. The average wind speed for this period was2.19 m s1.mesoscale advection.The Bowen ratio is typically low (<0.5), particularly at Site 3 where it remains lessthan 0.3. The park sites show an increase in j3 throughout the day again indicating thelack of warm air advection over the park. Rather QH is an energy sink for most of theday.5.2.3 The influence of fetch on QEFetch has an influence on evaporation in Trafalgar Park. As outlined in Chapter 3,Section 3.1.2, it is likely that this pattern reflects both the soil moisture and turbulenceregimes in the park. Data for the daytime QE on both July 24 and July 25, when thewind was from the southwest sector are used for this analysis (Fig. 5.9). This is a typicalwind direction so cumulative effects will show up in soil moisture.While no data were collected on the turbulence regime in the park, a likely scenario111U —+ Quiet zone TurbulentwakezoneVReadjustmentzone-Qz = Quiet zone shed by small step change.Figure 5.10: Possible scenario for the turbulence regime in Trafalgar Park, Vancouver.can be hypothesized given the discussion in Chapter 3, Section 3.1.2. In Trafalgar Park,with a southwesterly wind direction it is likely that the flow is disrupted by two features atdifferent scales. The first is the school buildings which are approximately 10 m high, 44 min the along—wind--direction and 110 m wide (perpendicular to the flow) and separatedfrom the park by a gravel field about 78 m wide. The wake shed by this complex islikely to extend into the park with a quiet zone mainly located over the gravel field,but possibly affecting Site 1. The turbulent wake zone shed by the buildings probablyinfluences Sites 1—3, while Site 5 may be in the readjustment zone. The second featurecausing flow distortion is the 0.8 m step change from the gravel field to the park. This islikely to effect both Site 1, which is in a quiet zone, and perhaps Site 2, which may be ina readjustment zone from this smaller wake. Given these two features a likely scenariocan be sketched for the overall turbulence regime of the park (Fig. 5.10).Although spatially variable, soil moisture shows an increase with distance from theupwind park edge (Fig. 5.11). This indicates that microscale advection is drying theschool complex gravel field 4 parkDistance112soil near the leading edge. Despite this general increase there is considerable variabilityover small scales. This is partly due to the installation of tile drainage which resultsin alternating bands of drier and more moist soil (this was evident in late August afterseveral weeks of drying). Also the turbulence regime will impact the trends in soilmoisture. Site 1 which is likely in a quiet zone (and hence sheltered from drying), hasa soil moisture content of 26%. Site 2 with a low moisture content of about 18% ismore exposed (possibly affected by two wakes: at the larger scale, it is probably in theturbulent wake zone shed by the school buildings, however it is also in the readjustmentzone shed by the step change in elevation near Site 1). Sites 3 and 5 which are beyond theinfluence of the smaller wake, have moisture contents of about 25% and 32% respectively.However, given that Site 3 may be in a turbulent wake shed by the school, it is surprisingthat the moisture content is not lower. Towards the far downwind park edge the moisturecontent is lower probably due to micro-advection from the adjacent road drying out thesoil.The impacts of both the turbulence and soil moisture regimes are hypothesized asfollows. Site 1 is in a quiet zone and is more decoupled from the flow, with a lower QEdue to reduced eddy diffusivity. Site 2 is affected by two wakes, the smaller of which maycounteract the higher diffusivities expected at this Site. Site 3 is in a turbulent wakezone with better coupling to the flow and consequently has an increased eddy diffusivityand hence higher QE. Site 5 is likely in a readjustment zone with intermediate couplingbetween Sites 1 and 3. Hence eddy diffusivity is decreased, resulting in a lower QE. Thisscenario is a possible explanation of the observed trend in QE across the park (Fig. 5.9).However without turbulence measurements within the park it remains speculation only.113405 ++ -I:$- 30G) 3 +1: 2 + +00 20 40 60 80 100 120 140 160 180 200Fetch (m)Figure 5.11: Average soil moisture for the observation period in Trafalgar Park, Vancouver.5.2.4 Summary of the energetics of a suburban park in a cool summerMediterranean climateThis analysis of the spatial variability of the surface energy balance in Trafalgar Parkoccurs over a drying phase. In the moist park, the SEB is dominated by evaporationwhich uses about 70—80% of net radiation. A strong edge effect with elevated evaporationis not observed. Rather, with distance from the upwind edge, evaporation increases upto about 25 m and then decreases to 100 m. This pattern probably reflects the turbulentregime in the park, due to obstacles at the upwind edge. Sensible heat accounts for12—20% of net radiation, and may be negative early in the morning and again for a fewhours in the late afternoon. This allows evaporation to exceed Q*. However, for most ofthe day QH acts as an energy sink, therefore the park has little propensity to exhibit edgeand oasis advective effects. The ground heat flux is small, using 8—13% of net radiation.1145.3 The urban park in its context — park, suburban and rural comparisons5.3.1 Energy partitioning in different land—use typesThis section discusses the differences in energy partitioning between the moist Trafalgar Park and suburban and rural (bush—bean crop) sites. Four days, July 25—28, 1992(YD 207—210) are selected for comparison with synchronous measurements at all sites(Fig. 5.12 and Table 5.2). Data from Site 5 in the centre of Trafalgar Park are used in thiscomparison. The energy partitioning at this site is very similar to that observed in the3—day average for YD 205—207 but there is less variability given the more settled synopticconditions. Given that Trafalgar Park was not irrigated (as was desired), data from astudy by Oke (1979) of evaporation from an irrigated suburban lawn in Vancouver, areincluded for comparison.The surface energy balance of the park is most similar to the rural site. Inter—sitedifferences of net—radiation are small (<5% daily and < 10% daytime). Previous studieshave shown net radiation at this suburban site exceeds that at a grassed rural site byabout 4% (Oke and McCaughey, 1983; Cleugh and Oke, 1986) i.e. comparable to thedifference in daily net radiation between the suburban and park sites. While both thepark and rural sites receive similar net radiation and are dominated by the latent heatflux, evapotranspiration at the park greatly exceeds that at the rural site. At bothsites QE exceeds Q* for a short period in the late afternoon. The sensible heat flux isrelatively low at both these sites but especially in the park. The ground heat flux is ofminor importance at both sites. On the other hand the suburban SEB is dominated byQH with the storage flux playing an important role in the morning.The average for the turbulent fluxes indicates the dominance of evaporation at thepark and rural sites. The daily ratio of evaporation in the park compared to the suburban rate is 2.25 which is lower than the 3.09 observed by Oke (1979) in a comparison of115a)Suburban b)Park600500‘ 400E3001;’ >‘&5 200-o-ox x 1002>s > 01) i5L t-100—200c) Rural0,C0)a)CLIFigure 5.12: Surface energy balances for the suburban, park and rural sites in Vancouver.Data are four—day averages. On an hourly basis net radiation is within 10% of its average,while the turbulent and storage fluxes may vary up to 20—25%.116LAT LATLATTable 5.2: Daily and daytime (Q* > 0) energy balance fluxes and and ratios for sites inVancouver. All fluxes are four—day averages in W m2.site Q* QE QH QG Qsa) DAILYSuburban 140 44 98 -2 0.31 0.70 2.25 0.42Park 135 99 32 3 0.74 0.24 0.33 1.08Rural 143 82 56 4 0.58 0.39 0.68 0.79Park/Suburban 0.96 2.25 0.33Park/Rural 0.94 1.21 0.57Suburban’ 156 53 72 31 0.34 0.46 1.34 0.60Lawn’ 95 100 -13 8 1.05 -0.13 -0.13 1.31Lawn/Suburban2 1.00 3.09 -0.30b) DAYTIMESuburban 314 74 181 59 0.23 0.58 2.46 0.44Park 282 196 61 24 0.70 0.22 0.31 1.07Rural 294 145 114 34 0.49 0.39 0.79 0.84Park/Suburban 0.89 2.65 0.34Park/Rural 0.96 1.35 0.54‘Oke, 19792To avoid complications from shading, these values are scaled.117evaporation from the lawn and that over the same suburban site. The park site and thelawn from the Oke (1979) study have evaporation rates in excess of equilibrium evaporation but only the irrigated lawn is significantly influenced by sensible heat advection.At the rural site the average daily c is 0.79 indicating the crop is moisture limited. Thiscrop was not irrigated and received less precipitation (6.6 mm) than the park on YD 203.Consequently QE is less than Qq.A comparison of the flux ratios at each site, clearly demonstrates the different energypartitioning between the land—use types (Table 5.2 and Fig. 5.13). At the park 74% ofdaily net radiation is directed into evaporation; this is considerably lower than the 105%observed by Oke (1979) for an irrigated lawn. Again this reflects advection of warmair being a more important energy source for the SEB of the lawn when soil water isartificially maintained near field capacity. At the unirrigated rural site only 58% of theavailable energy is channelled into QE. At the suburban site only 31% of Q* is used inevaporation. For the duration of this observation period a sprinkling ban for lawns wasin effect resulting in lower external water use by residents. Hence QE and QE/Q* forthis suburban site are lower than previously observed. In years without an irrigation bana more typical value is closer to 40%. At all sites the ratio QE/Q* shows an increasethroughout the day particularly in the late afternoon as sensible heat at the park andrural sites, and storage at the suburban site, turn negative i.e. act as energy sources.Due to the larger amount of energy channelled into QE at the park and rural sites,the ratio QH/Q* is much lower and has a corresponding decrease in the late afternoon.At the suburban site, due to the drier surfaces, QH is the dominant flux and the ratioQH/Q* is much higher and increases throughout the day to as great as 80%.The ground heat flux is of minor importance in the park and has a small role at therural site whereas the storage heat flux at the suburban site is substantial, accountingfor 60% of Q* in the morning. Roth (1991) observed a similarly sharp increase in the1181.0 — -500C 2rC1 .,0500 700a) QE/Q b) QH/QCC•1’LuburnCr0Pork •.v ‘.VV700 900 1100 1300 1500 1700 1900LATc) Bowen ratio43900 1100 1300 1500 1700 1900LAT500 700 900 1100 1300 1500 1700 1900LATFigure 5.13: Daytime fluxes normalized by net radiation for the suburban, park and ruralsites in Vancouver. Data are four—day averages.119morning and then an approximately constant rate until afternoon. As QH becomes moredominant throughout the day, iQ3 declines and is negative by 1600 LAT.In the park the Bowen ratio is fairly constant at about 0.36 confirming the dominanceof the latent heat flux in the SEB of the park. However as it remains positive there isno evidence of enhanced mesoscale sensible heat advection. In contrast, Oke (1979)observed a Bowen ratio of —0.13 over an irrigated lawn, clearly indicating an edge effect.Why then does the park not exhibit a tendency to behave as an oasis? The sandysoils and tile—draining in the park may have affected the rate of recovery from recentrain. This drainage regime, together with the lack of irrigation, meant that the parksoils were relatively dry resulting in a warmer surface which reduces the temperaturedifferential between the park and urban surfaces. This surface temperature forcing wasclearly illustrated in the thermal images of Vancouver parks which showed substantialdaytime surface temperature differences between irrigated parks and surrounding urbansurfaces while unirrigated parks have temperatures more similar to those of the urbansurface (Section 2.2.2). Hence for unirrigated parks with free draining soils, there is lesspropensity for warm air advection that could lead to elevated evaporation rates and edgeand oasis—type behaviour.At the rural site the Bowen ratio is greater than one in the morning reflecting thedominant role of sensible heat at this time. The ratio then declines throughout the dayto 0.35 at 1700 LAT. At the suburban site, due to the dominance of the sensible heatflux, the Bowen ratio is greater than two for most of the day. This ratio is much higherthan that observed in previous studies at this site due to the unusually dry suburbansurface resulting from the sprinkling ban.1205.3.2 Summary of the urban park SEB compared to other land—use typesin a cool summer Mediterranean climateThe unirrigated urban park is dominated by the latent heat flux and has a daily averageQE 20% higher than the rural site and more than twice that of the suburban site. Sensible heat is of secondary importance and there is little enhanced advective influence onevapotranspiration. Consequently the park does not behave as an oasis. It is suggestedthis is due to the free—draining park soils and a lack of irrigation. The ground heat fluxhas a minor role.At the rural site the SEB is again dominated by QE, but QH is more important hereaccounting for 39% of Q*. Like the park, QG has a small role. At the suburban siteQH dominates the SEB using 70% of Q*. Evaporation is particularly low for this sitereflecting lower external water use by the neighbourhood under a sprinkling ban. On adaily basis the storage flux is of minor importance however it is a major component on anhourly basis acting as a sink early in the morning and as a source in the late afternoon.Although the park was not irrigated, there was heavy rainfall four days prior to thisaveraging period. Most of the park drying occurred in the four days after the rain. Forthe duration of the averaging period, evaporation rates had levelled out and daily totalswere very similar. This transition phase of the park limits the generality of the results.However, the results do indicate the energetics of a moist park. Similarly, the SEB forthe suburban site is also limited in its generality given the unusual case of a city—wideirrigation ban.5.4 Comparison of park energetics in different climatesThe original experimental design aimed to compare the SEBs of parks in different climatesto determine the influence of background climate on energy partitioning. Unfortunately121a) Surface temperature difference b) Vapour pressure difference171513C)119Figure 5.14: Comparison of surface forcing on evaporation between the Sacramento andVancouver parks and their suburban environs. The surface temperature difference between urban and park surfaces are in (a), while (b) is the vapour pressure differencebetween the near—surface park air and the overlying air.the Vancouver study coincided with a city—wide irrigation ban so parks were not irrigated.As a result water status as well as climatic regime differs between the parks. The dataused in this comparison are the two—day averages from Orville Wright Park in Sacramentoand the four—day averages (for the settled synoptic period, YD 207—210) from TrafalgarPark in Vancouver. In the absence of an irrigated park, data from Oke’s (1979) study ofthe SEB of an irrigated lawn in Vancouver, are used to provide a first order comparison.A comparison of the surface temperature differences between the parks and nearbypaved surfaces in both Vancouver and Sacramento (Fig. 5.14a) shows a much largerdifference in Sacramento. The wetter (and hence cooler) park surface in Sacramentoincreases the park—urban surface temperature difference. Similarly, the vapour pressuredifference between the near—surface park air and the overlying air is also much greaterin Sacramento (Fig. 5.14b).LAT LAT122A summary of the daily SEB fluxes and ratios of fluxes for Orville Wright and Trafalgar Parks and the irrigated Vancouver lawn is given in Table 5.3. The park SEB fluxes,normalized by their daytime maxima to aid examination of their temporal behaviour inthe different locales, are presented in Figure 5.15.Table 5.3: Comparison of daily energy balance fluxes (W m2), and ratios for the Sacramento park (Orville—Wright), Vancouver park (Trafalgar) and an irrigated Vancouverlawn.Surface Q* QE QH QG /3Vancouver park 135 99 32 3 0.74 0.24 0.33 1.08Vancouver lawn 95 100 —13 8 1.05 —0.13 —0.13 1.31Sacramento park 119 142 -16 —7 1.19 —0.13 —0.11 1.52Vancouver park/suburban 2.25 0.33Vancouver lawn/suburban 3.09 -0.30Sacramento park/suburban 3.16 -0.22The energetics of the two parks are very different. Vancouver, at a higher latitude, hasa slightly longer daylength resulting in greater radiant energy input (Table 5.3). Bothhave negligible ground heat fluxes. This small flux is lagged in the Sacramento parkdue to the thick grass canopy. QE shows a similar temporal pattern in both parks, butthe ratio QE/Q* reveals that a much higher proportion of net radiation is directed intoevaporation in the Sacramento case. The irrigated Vancouver lawn has a SEB similar tothat of the Sacramento park with QE exceeding Q* on a daily basis. At these irrigatedsites evaporation exceeds the potential rate and is thrice that at nearby suburban sites. Incontrast, the moist Vancouver Park has a daily evaporation rate well below its potential.Both irrigated surfaces have the same small negative daily ratio of QH/Q* whereas123(a) Q/Q. max (b) OE/OE maxFigure 5.15: Comparison of park SEB fluxes normalized by their maxima (i.e.Q,/Qmax). Data are ensemble averages from the Sacramento and Vancouver studies. The normalizing maxima values (in W m2) for Orville Wright Park, Sacramento;:c0=Q529 1j=Q435 QI’ = liO,Q = 24; and for Trafalgar Park, Vancouver:1241 .21 .00.80.6Eb 0.400.20.0—0.21 .—0.6—0.8—1.0—1.2—1.40 500 1000 1500 2000LAT(c) OH/OH max+A*+..’—*—— Sacramento.Vancouver1 . . 0.10o —0.1-0.3—0.5—0.7—0.9—1.14;—— SacramentoVancouver44.f 414/A--‘--. A.c’0 500 1000 1500 2000LAT(d) Q/Q max500 1000 1500 2000LAT500 1000 1500 2000LATthe Vancouver park has a small but positive ratio. This highlights the difference inenergetics between the two parks. In the unirrigated Vancouver park the soil is muchdrier, particularly near the upwind edge where microscale—advection has depleted soilmoisture beyond that of the rest of the park (section 5.2.3). Consequently the evaporativeforcing is lower than expected for an irrigated park and hence the park has a positive QHthrough most of the day (it is negative for only two hours in the evening). With distanceinto the park evaporation varies in response to the turbulence regime (an initial increasecoinciding with a turbulent wake zone and decrease in the readjustment zone (Fig. 5.9and 5.10).In contrast, sensible heat fluxes over the irrigated surfaces (both the Sacramento parkand Vancouver lawn) are negative in mid—afternoon, which coincides with the time ofmaximum evaporative forcing (Fig. 5.14). Thus evaporation from these well—watered surfaces is substantially assisted by advection. In Sacramento both micro— and mesoscaleadvection are important in the afternoon and evening and sustain high rates of evaporation. In response QE shows an exponential decrease with distance into the park.The irrigated Vancouver lawn is small (ca. 75 m2) so microscale advection is likely tohave been the most important type involved. Over a more extensive irrigated surface,oasis—type advection might also occur in Vancouver.In moist or wet urban parks evaporation dominates the SEB and evaporation canexceed the potential rate if surface forcing is sufficient to induce edge and oasis—typeadvection. The free availability of water, which in summer implies irrigation, appearsto be a criterion in order for QE to exceed QEp. In addition there must be a sufficientpark—suburban temperature and vapour pressure deficit differential. Given the similarityof the energetics of irrigated surfaces in different climates, it is suggested that soil moisture, rather than background climate, is of greatest importance in determining energypartitioning in parks.125Chapter 6Scale Model Design for Nocturnal Cooling in Urban ParksThis part of the research aims to gain a better understanding of the controls on parkcooling following sunset. A methodology is developed for scale modelling of nocturnalcooling in urban parks.6.1 Considerations in nocturnal park coolingThe descriptive survey of air and surface temperatures of parks (Chapter 2), showedthat except for open, dry grass types, parks are generally cooler than the surroundingneighbourhood throughout the day. For parks with large shade trees the urban—parktemperature differential may reach a maximum in the late afternoon. However for mostparks, the PCI is best developed at night, soon after sunset for garden and savannahparks, and near sunrise for open grassed parks.There are several reasons why a park may be a “cool island” relative to the surrounding neighbourhood. By day, the park may have less radiant energy input and/or itmay use more of its heat in evapotranspiration or heat conduction to the ground. Theturbulent sensible heat transfer and the availability of anthropogenic heat input however,are likely to be less than for the surrounding city. These factors contribute to the parkheating up more slowly. At night, in addition to these factors, the park may have greaterradiative cooling than the city due to its surfaces possessing higher sky view factors. Theresearch in Sacramento and Vancouver parks (Chapters 4 and 5) provides comparisonsof energy balance partitioning in parks and suburban neighbourhoods, and can assist in126determining which of these processes are likely to be important.For open grassed parks, it is possible for the net radiation to be higher or lower thanthe suburban value, depending on the albedo of both surfaces. If the park is well treed,net radiation at the surface will be lower because of shade and this will have an importantinfluence by slowing the warming of the park by day, and reducing heat loss at night.Heat conduction into the ground is much less in the park than in the suburban neighbourhood. Comparison of the two showed that during daytime, the park in Sacramentohad about 85% less while in Vancouver the park had 60% less heat conducted to storagethan the city (p.90 and p.117). By night the heat release is larger in the city than fromthe park.Convective sensible heat transfer is much lower in parks than in the surrounding.city, and in fact in some cases the park may act as a sink for QH rather than a source.This occurs when the park acts as an “oasis”, often resulting in QE exceeding Q*, anda downward flow of sensible heat acting as an extra energy source. This oasis—typebehaviour was exhibited in the well—irrigated Orville Wright Park in Sacramento. InVancouver, probably due to the lack of irrigation, Trafalgar Park did not act as anoasis, however sensible heat in the park was about 70% lower than in the suburbanneighbourhood.By day convective latent heat transfer has the potential to be much higher in apark than in its surrounding city. Certainly in both the Sacramento and Vancouverparks, the latent heat flux from the park is substantially higher than the suburban value.The magnitude of the park—city difference in QE depends upon many factors including:soil moisture status of the park relative to the city, the macroscale climate regime, thesuburban density and local gardening practices. On the other hand at night QE continuesat a much reduced rate. By sunset, high rates of evapotranspiration in some well—irrigated parks may establish the park as a cool island. However, at night under calm,127clear conditions when urban—park differences can be large, turbulence is suppressed.This suggests that nocturnal cooling of parks must be due primarily to radiative andconductive processes accentuating the PCI which is already present at sunset.6.2 Objectives of scale modellingThe aim of the scale modelling portion of this study is to determine the relative magnitudeof the controls on nocturnal park cooling. Following the discussion in section 6.1, theobjectives are to assess:• the role of radiative transfer;• the role of conductive transfer;• the role of latent heat transferin nocturnal cooling. Consideration is restricted to the case of calm and cloudless weatheras the park effect is greatest under these conditions. The model simulates the effects ofurban and park surface geometry, thermal properties and evapotranspiration.6.3 Scaling considerationsIt is important to state the extent to which model:full scale similitude is preserved.The model simulates cooling after sunset, therefore it is necessary to scale the processesgoverning heat loss as well as the physical properties of the surfaces which govern theability to release heat.The main nocturnal processes are radiation and conduction and the relevant surfaceproperties are radiation geometry and thermal admittance. It is not necessary to scale128the radiative process since all radiative transmission is at the same speed and the characteristic length scale is of negligible dimensions (Oke, 1981). The surface geometry islength scaled.The model to simulate cooling by conductive transfer manipulates the thermal properties of the “park” relative to the “city”. The appropriate property is thermal admittance(p) which is related to the thermal conductivity (k) and heat capacity (C) of the surfaceby:(6.1)It is a measure of the thermal response of the surface to the heat flux across the interface.With low values of thermal admittance a surface will readily accept or release heat, butconversely for high p heat is accepted or released slowly. Surfaces of high p exhibit asmaller amplitude temperature wave than those of low p. The p value for urban parksmay be higher or lower than the city depending mainly on the relative differences insoil moisture content and the nature of the urban fabric. Rather than attempting toapproximate real world thermal admittances of park and urban surfaces, the park isconstructed of materials that have either higher or lower thermal admittances than themodel urban surface. This approach provides a first approximation to the influence onthe park cool island, PCI, of the thermal admittance differences between the park andthe city.Scaling evapotranspiration is a complex problem. Simple models of leaves and plantcanopies have been used to simulate the flow field and heat and mass transfer (e.g.Schuepp 1972, 1973; Chen et al., 1988a,b; Siegner et al., 1976; Coppin et al., 1986).However the problems surrounding this reductionist approach are immense. Jarvis andMcNaughton (1986) and Baldocchi (1989) caution against merely extrapolating from onelevel to the next and emphasize the central importance of the scale of the phenomenon129(leaf, plant, canopy, region). Due to important links between plant response and theenvironment, it is unlikely that physical scale modelling can be applied to study evapotranspiration from a park surface, particularly if results are to be generalized.Given these difficulties, the “park” is treated as a moist surface with no biologicalcontrol but, by using an appropriate surface cover, some resistance is provided to freeevaporation. Given that plants generally close their stoma at night this approximation,which neglects plant physiological control, is considered acceptable.6.4 Scale modelsTo study each park cooling mechanism, three different models are implemented. Whilesuch an approach is inherently reductionist and therefore over—simplifies the park system,it gives a first order approximation of the magnitude and extent of cooling to be expectedif each mechanism were operating in isolation. Once each mechanism has been studied inisolation, the combined effects of each mechanism are determined through a combinationof model designs.The models are based on hardware scale representations of a “park” surrounded byan “urban environment”. The model consists of a base of wood (0.83 mx0.83 m, 37 mmthick Douglas fir) upon which blocks of the same wood (to simulate “buildings”) arearranged into canyons. The middle of the model is the “park”, which is simulated invarious ways depending on the mechanism being investigated. The base is insulatedfrom below by a layer of polystyrene.A nocturnal cooling cycle is simulated in the following manner. The model is irradiated by nine 500 W halogen lamps. The length of heating is determined by scalingconsiderations. The thermal diffusivity, ,c, for the fir base was initially estimated as9.7x108m2 s (Chapman, 1984). The penetration of a diurnal heat wave into the soil130is represented by:T(z, t) = + A0e’ sin (wt—(6.2)where T is temperature; z is depth; t is time; T is the average surface temperature; A0is the amplitude at the surface; D is the damping depth; and is the angular frequencyof the oscillation ( = 2ic/D). An appropriate damping depth is chosen to regulate thediurnal heating and cooling cycle. Given that the model base is 37 mm thick, a dampingdepth of 10 mm was arbitrarily chosen. This results in: .‘ = 0.0105 s_i which gives adiurnal period, P = 27r/w=3239 s or 54 minutes.In the real world, soil temperature is approximately isothermal from the surface tothe damping depth shortly before sunset. This was achieved by turning the lamps on andoff so as to propagate a series of small heat waves. This condition of isothermality hasbeen used as a boundary condition to initiate cooling (Brunt, 1941; Oke, 1981) in priornumerical and scale models of nocturnal cooling at the surface. This boundary conditionis hereafter referred to as the model “sunset”. At “sunset” the model is quickly removedfrom beneath the lamps, a polyethylene tent is placed over it, and it is moved into a smallroom at a temperature of about 20° C. To create a surface to sky temperature differentialfor the model that is comparable to that in the real world, the model is heated to beisothermal at about 60° C. This establishes a temperature differential of approximately40°C between the hot model and the cold “sky” (the ceiling and walls of the room)at room temperature. This temperature differential allows observation of about 15—20°C cooling over a ten minute period. The polyethylene tent avoids convective transferbetween the hot model and the air in the room. The surface cooling is monitored forabout ten minutes (625 s) following sunset. This corresponds to the time for the modelto approach the maximum amount of cooling given the rationale for scaling time. Thisis slightly less than initially estimated from the time scaling scheme (for a period of13154 minutes the time from a condition of isothermality to the minimum temperature isabout 16 minutes. However, the thermal diffusivity of the model base was subsequentlydetermined by differential scanning calorimetry to be 5.25 x i0 m2 s, hence the revisedtime for cooling.6.4.1 Simulation of radiative coolingThe model to simulate radiative cooling involves altering the dimensions of the surfacegeometry to simulate different types of urban park (Fig. 6.1). The “buildings” are arranged in canyons with a height to width ratio of 1 : 1 to simulate the urban environment.The simplest case of “park” is bare wood for an open grassed park. For the case of parkswith trees, model “trees” fashioned from foam rubber are placed in the park to mimicparks with tree borders, savannah parks and garden parks. A more—or—less continuouscanopy of foam is placed over the park to simulate a forested park. The dimensions ofthe surface geometry are scaled to give a 1 625 relation between the model and thereal world. Hence “buildings” 40 mm high in the model, represent about 25 m (or 5storey) buildings in the real world, and the 40—100 mm high “trees” both conical andbroad shaped (spherical) in the model represent 25—62.5 m tall trees in the real world.Two sizes of park are simulated. The first is a small 1.6 ha park, approximately 0.20by 0.20 m in the model; the second is a 5 ha park, approximately 0.36 by 0.36 m in themodel. To maintain a constant thermal mass of the model overall, the blocks removedwhen changing park size, are stored as part of the model base.6.4.2 Simulation of conductive coolingThe simulation of conductive cooling differences is achieved through manipulation of thethermal properties of the park relative to the city. The background “city” is alwaysconstructed of wood (fir) while the “park” is made of different materials (Fig. 6.2). The132(a) Grass park with tree border (b) Savannah parkFigure 6.1: Scale model design to simulate radiative cooling in different park types. The“grass park” is open with no “trees”. The park dimensions are O.36x0.36 m and thebuildings are 40 mm high and wide.(c) Garden park (d) Forest parkA133materials used together with their estimated thermal admittance are given in Table 6.1.The thermal conductivity and specific heat of the fir was determined through differentialscanning calorimetry. Both properties are temperature dependent. The following average values appropriate for the range of temperatures (20—70° C) in this study are used:0.265 W m’ K—’ for thermal conductivity and 1122 J kg’ K—’ for specific heat (c3).With knowledge of the density (p) the volumetric heat capacity (C) is derived:C = pc3, (6.3)and thence thermal admittance (p) from equation 6.1. Thermal diffusivity (ic) can alsobe calculated:(k)2 (6.4)The thermal admittances of the remaining materials in Table 6.1 are estimated by anumerical model (see Section 6.6.2) which is validated against the measured Douglas fir.Table 6.1: Thermal admittance of the materials used in scale modellingMaterial(3 m_2s4 K—’)Douglas fir 366Plywood 234Oriented—strand—board (OSB) 198Low density fibre—board (LDF) 147Concrete 7276.4.3 Simulation of evaporative coolingThis model set—up involves a moist “park” surrounded by a dry “city”. The model has afir “city” as before with the park section replaced by a moist surface (Fig. 6.3) consisting134insulationpark (different material)/“city” (fir)Figure 6.2: Scale model design to simulate conductive cooling in urban parks.initially of a pan of water, and in a subsequent run, by moist blotting paper. In thiscase the model is heated without the moist park, until the temperature at the 10 mmdepth is approximately 60° C. The wet “park” is then placed on the model and after afurther period of heating, sunset is initiated and the cooling of the surface is monitored.At sunset the vapour pressure deficit is typically about 1700 Pa.6.4.4 Combined influences of geometry, thermal admittance and evaporationThe combined influences of surface geometry, thermal admittance and evaporation areassessed in a further set of experiments. Firstly, surface geometry and thermal admittanceare combined by adding canyon geometry to the city together with materials of differentthermal properties for the park. The addition of surface geometry provides view factoreffects as well as changing the thermal admittance of the surface.Secondly, the combined effects of thermal admittance and evaporation are examined0.83 m135“park’ (nlois surface)“city” (fir)insulationFigure 6.3: Scale model design to simulate evaporative cooling in urban parks.in experiments using a flat model with a moist park. However, unlike the parks used inthe evaporation experiments, the same materials used in the thermal admittance experiments are soaked for several hours and then inset as a park. This creates both thermaladmittance and evaporative differences between the park and the city.Finally, experiments are run with the combined influences of surface geometry, thermal admittance and evaporation. This is set up as described above for the thermaladmittance and evaporation experiments except for the addition of surface geometry onthe city. This analysis also examines the behaviour of each park material under differentboundary conditions: flat, dry; flat, wet; and for models with urban canyons surroundinga central park — canyons, dry; and canyons, wet.0.83 rn1366.5 Measurements6.5.1 Surface temperature measurement by thermocouplesThe surface temperatures of each model are sampled by an array of thermocouples extending from the edge of the “city” to the centre of the “park”, as well as in “urbancanyons”. Thermocouples are also installed to sample temperatures of different facets ofthe city (ground, canyon wall, canyon roof), of tree canopies (when present) and specialspatial coverage across the different park set—ups. Thermocouples also measure “ground”temperatures at various depths (10 mm, 20 mm and 37 mm).The sensors are 36 awg, copper—constantan thermocouples insulated with teflon coating. They were tested prior to use on the model to determine the inter—sensor precision.The thermocouples were placed in a sealed box in a darkened laboratory for a period of12 hours. Temperatures were sampled every 30 s and averaged over ten minutes. After12 hours the standard deviation of the measured temperatures was less than 0.4°C.The thermocouples are attached to the surface following the method of Fairey andKalaghchy (1982). This involves the construction of small arcs (approximately 5 mm)between the junction and the insulated wire. The arcs are attached to the surface using athin layer of tape rather than adhesive to allow the thermocouple positions to be moved.Seven of the thermocouples in the transect from the city to the centre of the parkare sampled at 0.5 Hz while the remaining 13 thermocouples are sampled at 0.1 Hzusing a multiplexer. The data are continuously measured on a CSI CR21X data logger.The standard error in temperature measurements is estimated to be 1.03° C. The mainsources of error occurring are attributed to the reference temperature within the 21X,thermocouple output and mounting errors (Appendix B). Tests of replicability show theerrors are much smaller — typically less than 0.5° C (Section 6.6.5).1376.5.2 Remotely sensed surface temperature measurementsAn AGEMA Thermovision scanner is used to remotely sense model surface temperatures.The scanner is placed approximately 1.12 m away from the cooling model, looking downat the model at an angle of about 300 from the horizontal. With the 12° field of view, thisresults in a viewed area on the model of 0.137 m2. This is sufficient to encompass an entiresmall park and the adjacent city canyon, or half of the large park and its adjacent canyon.The polyethylene tent is extended to the AGEMA lens via a sleeve, so that the scanneris looking directly at the model, hence the transmissivity is approximately 1. Images arerecorded at 0.1 Hz. For model runs with a moist park, surface temperatures are digitizedfrom the AGEMA images as the thermocouples do not have good thermal contact whenthe surface is damp. For the range of temperatures encountered in the modelling, theAGEMA can estimate surface temperature to within 0.5°C (Section 2.1.2).Figure 6.4 shows a comparison of surface temperature measured by the thermocoupleand AGEMA methods. The systems are well correlated but the AGEMA consistentlyestimates higher surface temperatures than the thermocouples; particularly at temperatures greater than 50° C. This is not of great concern. It is thought that the differencemay be mainly due to mounting errors for the thermocouples (approximately 0.4°C).6.6 Analysis6.6.1 CorrectionsMeasurement errors are expected, for example a thermocouple may lift slightly duringthe run thus giving a lower (air) temperature rather than a true surface temperature.Comparison of the temperatures measured by the two systems is used to detect andcorrect such a problem. This should require consideration of an emissivity correction tothe AGEMA temperatures because a surface emissivity of either unity or 0.97 is assumed13858C)0-4-,q)E-4-,LUC)56545250484644424040 42 44 46 48 50 52 54 56 58Thermocouple temperature (°C)Figure 6.4: Comparison of surface temperature estimates from the AGEMA and thermocouples. Data are averages for three model runs.139during measurement. The need for correction was checked as follows.The emissivity of the model materials was determined using the method of Davieset al. (1971). This entailed placing a cone with a polished mylar surface over thematerial to allow the surface to behave as a blackbody. The temperature of the surfacewas measured before the cone placement by an Everest 4000A infra—red thermometerto give the apparent surface temperature (Tr). Immediately after placing the cone, thetrue surface temperature (Ta) was measured using an infrared thermometer viewing thesurface through an aperture in the polished cone. The radiative sky temperature (Tk)was measured at several zenith angles and azimuths and averaged. The emissivity of thesurface () was estimated:(6.5)Table 6.2 shows the estimated values of emissivity for the model materials. The rootmean square error (RMSE) for estimates of emissivity is 0.02 (Appendix B). Sensitivitytests were run to estimate the error incurred by using an assumed of 0.97 or 1.0 by theAGEMA system (Fig. 6.5). The following example illustrates the probable error involved.If an emissivity of unity is assumed for the fir, which actually has an emissivity of 0.95,the maximum error in the surface temperature is about 0.4°C at 70°C. The size of theerror decreases as the model cools. Given errors of this order, no corrections were madeto surface temperatures.6.6.2 Theoretical frameworkAssumptions of modellingThere are several assumptions both implicit and explicit in the modelling:• Cooling processes in the model are dominated by radiative and conductive processes, i.e the model design suppresses convective cooling.140(a) Emissvity of 0.97 assumed0.2—o-— 0.99•4.. 098-—0-- 0.970.1 0.960.95— — ——— —x—*0eq— 0.0I—•0l—0.230 40 50 60 70Temperature (°C)(b) Emissivity of 1 .0 assumed0.41.00.990.3 0.98 ..0.970.960 0.95 fl-..- ,,__x_‘—‘ 0.2..— 0—4-- .4_c01—0 --_e..G_4_.4-.4--A - A --.-a- . A0.0 - geese. 60.0 egg’30 40 50 60 70Temperature (°C)Figure 6.5: Sensitivity tests of the difference between assumed and actual emissivity onderived temperatures from AGEMA images: (a) shows the temperature difference whenc = 0.97 is assumed; (b) shows the temperature difference when = 1.0 is assumed.141Table 6.2: Estimated emissivities for model materials in both their dry and moist states.The means and standard deviations (u) are derived from three tests.Material c (dry) o (dry) £ (moist) a (moist)Fir 0.967 0.005Plywood 0.957 0.005 0.988 0.002OSB 0.953 0.005 0.986 0.001MDF 0.958 0.003 0.988 0.001LDF 0.963 0.003 0.989 0.002Concrete 0.969 0.004Blotting paper 0.959 0.002• As all radiative transmission is at the same speed and the characteristic lengthscale is of negligible dimensions, geometric scaling holds.• Cooling is proportional to the thermal admittance of model materials.• Time is scaled according to the thermal properties of the model materials andconsideration of the penetration of a sinusoidal heat wave into the base.• The surface—sky temperature differential at “sunset” is comparable to the realworld.• Anisotropy of the model materials is not considered a problem.Comparison of scale model cooling with predictions from a numerical modelTo determine if the model simulates real world conditions, it is tested against a numericalsurface heat island (SHIM) model developed by Johnson et al. (1991). This model142simulates nocturnal cooling of rural and urban surfaces under clear, calm nights and hasbeen carefully validated against field data. The model consists of a system of partialdifferential equations, hereafter referred to as the SHIMpDE approach. A full outline ofthe model is given by Johnson et al. (1991) and numerical details are given in Johnsonand Watson (1987, 1988). The urban and rural environments are modelled as layers inwhich conduction normal to the layer greatly exceeds conduction parallel to it. Heat flowwithin the layer is modelled by the one—dimensional heat conduction equation,ÔT .92T (6.6)where i is the thermal diffusivity (m2 s’) of the layer and T(x, 1) is the temperature atdistance x from the boundary at time t. Heat flow at the surface of a layer is inducedthrough radiative transfer and heat gain at a surface i, of absolute temperature T, whenexposed to other surfaces j (where j = 1, ..., N), of absolute temperature T, emissivitye,, and view factor is given by:N— (6.7)j=1where : is the emissivity of surface i and o is the Stefan—Boltzmann constant. Heat flowis modelled within N surface elements by:8Ti1,...,N, (6.8)oxwith the boundary condition for the interior of this layer given byT(O, t) = TG(t), (6.9)while the boundary condition for all exterior surfaces isk1(D,t) = L i = 1,...,N, (6.10)143whereN= i + 5L _u4(Dt)] (6.11)+iU — fk)’tIJ,kEjT4(Dj,t),k=1 j=1k1 jkis the net long—wave radiative flux density, using the diffuse—gray assumption. Thesubscripts G and s refer to deep soil, and sky respectively, while D, is thickness (m)and k, is the thermal conductivty (W m’ 1<—’) of element i and L .j. is the incominglong—wave radiation. The initial condition for each element isT1(,0) = f:(x) 0 x < D1 i = 1,...,N, (6.12)where f(x) are given temperature distributions.The model is initialized with both a surface and deep soil temperature, incominglong—wave radiation, the constants f, ,c, k and the view factors for each surface. Theview factors for each surface are calculated using analyses derived from Steyn and Lyons(1985) and Howell (1982), and also fisheye—lens photography with the method of Steyn(1990). Details are given in Appendix C. The model uses the Crank—Nicholson methodto solve equation 6.8 giving temperature of each surface as a function of time.The SHIMp approach has successfully simulated surface cooling (see Johnson et al.,1991) and is used in this analysis to determine if the scale model can produce realisticresults. An ensemble average of surface cooling data from five scale model runs withsimilar boundary conditions, is used to verify the model. Figure 6.6 shows a comparisonof cooling by the scale model with that predicted by the SHIMpDE model when initializedby scale model data. The numerical model slightly overestimates the cooling in the scalemodel at lower temperatures, i.e. the latter part of the “night”. However, agreementbetween observed and predicted results is generally good and verifies that the assumptions1446563*——— Observed61 - -- SHIMpo mode!(5) 59 -57Da5)535149(_f) -—47 - I-----II45.I I I I I0 100 200 300 400 500 600 700Time after sunset (s)Figure 6.6: Comparison of surface cooling observed in the scale model (flat fir surface)with that predicted by the SHIMpDE model when initialized by scale model data. Theobserved data are the averages of five model runs. Standard errors for the observationsare indicated.in the scale model are valid. Thus the scale model realistically simulates surface cooling.6.6.3 Derivation of thermal admittanceGiven the ability of the SHIMpDE model to simulate both full—scale and scale modelcooling, it is used as a tool to determine the thermal admittance of the remaining materials used in the scale model. This is necessary because it was not possible to have allthe materials laboratory tested to determine their thermal properties. The observed surface cooling from the scale model is compared iteratively to that predicted by SHIMpDEunder different trial values of k, ic and until good agreement is found.1456.6.4 Calculation of surface temperature park cool islands, PCI5sTo assess the impact of the chosen experimental variables on cooling, a park effect iscalculated for each model run. Despite attempts to evenly heat the model, there is alwayssome spatial variability of surface temperatures at “sunset”. To simplify matters spatialaverages of surface temperatures are used in the calculation of the park cooi island, PCI5.Both the average urban surface temperature, (averaged from mid—points of canyonfloors), and the average park temperature, T, are derived from three points. The PCI3is then calculated:PCI = — . (6.13)There may be a positive (or negative) PCI present at sunset so this offset is subtracted(or added) to set the PCI3 equal to zero at sunset. This is hereafter referred to as a“normalized park cool island,” NPCI3. The analysis compares the growth of NPCI3through the “night” between the different model set—ups.6.6.5 Replicability of resultsTo have confidence in the model, results must be replicable. For the simulations of radiative cooling in larger parks, each set—up was replicated at least three times. Comparisonshows that replication errors are generally small for park surface cooling, but errors arelarger for cooling in urban canyons. This is illustrated in Figure 6.7 which shows thecooling for urban canyon and park temperatures, and the resultant PCi and associatederrors for a “grassed park”. The cooling curve for the open park is relatively smooth andpossesses a low standard error. The curve for the canyon floor is more variable and hasa higher standard error. The variability may be due to convection (buoyant thermalsrising from the surface). While there are indications from field research that air withincanyons is unstable at night (e.g. Nakamura and Oke, 1988) this phenomenon would be146enhanced at the higher model temperatures. Thus the resulting standard error in a PCI3may approach about ±0.7°C. Given this variability in the urban reference, analyses ofthe PCI comment on the absolute cooling of the park surface.147(a) Average urban and park cooling0—2—4,— —6(-)—8—1000() —12—1 4—16—180 100 200 300 400 500 600 700Time after sunset (s)(b) Growth of the PCI after sunset0 100 200 300 400 500 600 700Time after sunset (s)Figure 6.7: Test of the replicability of the scale model to simulate cooling in a “grassedpark”. Average urban and park cooling curves (a) are shown together with the derivedPCI (b). The averages and their standard errors derived from four replicates.parkC)C-)z7654320—1148Chapter 7Scale Model Results of Nocturnal Cooling in Urban ParksThis chapter presents results from scale modelling of nocturnal cooling in urban parks.Results from the different model designs are given: surface geometry, thermal admittancedifferences, evaporative effects and combined effects. The combined influence of thesemechanisms is discussed in relation to cooling observed in real parks.7.1 Radiative cooling7.1.1 Influence of park sizeThe absolute size (dimension) of a park is an important control on its nocturnal climate.Close to the edge of the open grassed model park there is a sharp decrease in sky viewfactor, (Fig. 7.1). This edge effect is confined mainly to the perimeter of the park. Inthe case modelled, points from a quarter of the way into the park, to the centre, haveview factors greater than 0.85. These view factor effects are reflected in the growth of thePCI (Fig. 7.2). Near the edge of the large open grass park ( = 0.84) surface coolingis of a similar magnitude to that for the centre of the small park (& = 0.88). Cooling isgreater in the middle of a large park where is 0.96. However, for from 0.84—0.96,there is a similar rate of increase in PCI immediately after sunset. This suggests thatsmaller parks can attain a significant amount of cooling soon after sunset. To enhancecooling through the night, larger parks are more effective.The optimum park size mainly depends on the geometry of the urban surroundings.149By estimating for the centre of square open grassed parks of different sizes withconstant height of the buildings surrounding the park (see method in Appendix C, sectionC.1), the relation between and the ratio of park width height of the surrounds, isderived. As the ratio of park width to building height (or tree height if there is a treeborder) increases beyond about 7.5, there is little gain in terms of radiative cooling(Fig. 7.3). However as the park size increases a larger volume of air is cooled and thisincreases the potential for advection of cool air into the neighbourhood.7.1.2 Influence of park typePark type (defined Chapter 1, section 1.2.2) alters the configuration of vegetation andthis influences patterns of within the park. For the open grassed park there is a strongtemperature gradient near the park edge, but little difference between cooling at pointsin the central area of the park (Fig. 7.4a). The grassed park with the tree border exhibitssimilar cooling trends to the treeless park, except that cooling near the edge is especiallyweak, because the tree border reduces ‘çb5.Savannah and garden parks (Fig.7.4c,d) have more spatially—variable surface coolingpatterns due to their complex array of sky view factors. However, despite lower z/,3 forthe centre of these parks, they still display mid—park cooling comparable to that in opengrass parks. This may be due in part to greater surface forcing of cooling with a highersurface to “sky” temperature differential at the time of initalization (or “sunset”).The forest park (Fig. 7.4e) shows a great range of surface cooling environments. Thepoint on the “ground” a quarter of the way across the park cools little, because thesky is completely obscured. However, both the edge and the centre of the park exhibitconsiderable cooling, despite values of 0.31 and 0.28, respectively. This apparentanomaly may result from the sinking of cooler air from the top of the forest canopy ontothe “ground” surface. The top of the canopy has b3 about 1.0, and being made from1500mmbuiiiIgs50-1Figure 7.1: Map of sky view factor for a quadrant of the large open grass park model.The surrounding buildings are 40 mm high. The location of the thermocouples used inthe analysis is also shown.1 edge2 quarter of way into park3 centre4 cornerI—657595—A2parkA315100.-UEC060 5 10 15 20 25 30Ratio park width building heightTime after sunset (s)8765C).— 40ciz20Figure 7.2: The influence of park size on development of the PCI3 after sunset.1 . .Figure 7.3: Relation between the ratio park width: building height and sky view factorin the centre of the park. Building height was held constant, but can in practise, vary.152a) Grassed park—2—4—6cs —10C0aC)c) Savannah park—2—4—6—8—10—1 2—14—16—18—20b) Grassed park with tree borderd) Garden parkFigure 7.4: Influence of park type on cooling of different facets: (a) open grassed parks;(b) grass parks with tree borders; (c) savannah parks; (d) garden parks and (e) forestparks. Cooling rates in the urban canyon are shown for comparison. The t’ for eachfacet is given in brackets.153C)C00C)Time after sunset (s) lime after sunset (s)C-)a0C-)canyon floor (0A5)100 200 300 400 500 600Time after sunset (s)400 500 600Time after sunset (s)e) Forest parkC-)00C)—2—4—6C-)-.-- —8g -1008 -12—14—16—18—20zzz—canyon floor (0.45) — — -- --edgepork(0.31) — — — — —1/4waypark(0) —— ——raid park (0.28)0 100 200 300 400 500 600Time after sunset (s)foam has very low thermal admittance and almost no contact with the model base.A general sense of the cooling induced by different park types can be gained from(Fig. 7.5a—e). This shows the surface temperature distributions at the end of each model“night” (after 625 s of cooling). These images have not been normalized for any differencesin surface temperatures at sunset, so intercomparisons should be made with caution.Rather, they are used to illustrate spatial patterns. To aid comparison they are plottedwith a common temperature scale (and hence some resolution is lost).The effects of surface geometry are clearly visible. The edges of the park and urbancanyons are much warmer than other facets. In the more open parks (grass, and grass withtree border) the mid—park region shows greatest cooling. In the savannah and gardenparks, with interspersed trees, cooling is more patchy, and open spaces are distinctlycooler. Where trees are clustered surface temperatures are much higher. The treesthemselves are always much cooler (closer to air temperature) and in the forested parkthe extensive canopy provides an elevated cooling plane which is considerably cooler thanany of the surroundings.The resultant NPCIs at the end of the “night” for the different park types (Fig. 7.6a)range from small (2°C) for the forest park to large (7°C) for the open grass park (Table 7.1). The insulating canopy in the forested park slows “ground” surface coolingresulting in a small increase in the NPCI by the end of the night. However by “day”,this canopy shades the ground and can produce a substantial cool island which nocturnal cooling slowly accentuates. The increasing NPCI from open to forested park type isclearly related to greater (Fig. 7.6b).This effect of surface geometry is comparable to that observed by the numericalsimulation (surface heat island model, SHIM) of Oke et al. (1991) which was developedto illustrate the effect of surface geometry on urban heat island development. Theyfound that the presence of canyon geometry (H:W of 1.15) alone, resulted in a 4°C154(a) Open grass park (b) Grass park with tree border2’.-:;-—4. O.> CItriFigure 7.5: Park surface temperatures at the end of the model “night” (625 s). (a) opengrassed parks; (b) grass parks with tree borders; (c) savannah parks; (d) garden parksand (e) forest parks. 155(c) Savannah park (d) Garden park(e) Forest parka) Growth of the PCI87654320.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0Average 1bFigure 7.6: The influence of park type on: (a) the growth of the PCI3 (average b3 isshown in brackets) and (b) the relation between the average b3 and NPCI3.100 200 300 400 500 600Time after sunset (s)b) Relation between average and NPCI.C)C)3-z0—1—276,—‘ 5C-)C-)u_z32—0.0A156Table 7.1: The effect of park type on the magnitude of the average PCI. All PCI3s arein Celsius.Park type b3 PCI3 PCI8 at Growth/Decayat sunset end of night PCIgrass 0.92 -3.5 2.6 6.1grass with tree border 0.84 1.3 4.1 5.4savannah 0.82 -1.9 3.0 4.9garden 0.78 -1.3 2.8 4.1forest 0.20 4.8 7.2 2.4UHI development after sunset. However, they suggested that realistically, if the thermaladmittance of the rural area is sufficiently small, the maximum effect of geometry aloneis of the order of 7CC; i.e. similar to that observed here in the scale model of open grassparks.7.2 Thermal admittance effects on park coolingFor the case of the flat, dry park model the magnitude of mid—park cooling is inverselyproportional to the thermal admittance, t. The magnitude of the resultant PCI8s decreases with increasing 1a (Fig. 7.7 and Table 7.2). If ILpark >> Pcity (e.g. concrete) thereis a negative PCI at sunset and it steadily becomes more negative throughout the night.However, if IL park << itct (e.g. LDF) there is a positive PCI3 at sunset, and a modestgrowth through the night (less than 4°C). Materials with p closer to the background fir(plywood and OSB) show only a small (2°C) PCI3 after 625 s.Despite the small range of p used in this model design, the differences are largeenough to create an effect and indicate that in the real world, where these differences157o—2 °40 LDF— + OSB60 PLY0 100 200 300 400 500 600Time after sunset (s)Figure 7.7: PCI3s for open parks with differing thermal admittances.Table 7.2: The effect of surface thermal admittance differences (Lt) upon the PCI3. The“city” has no canyon geometry and has = 366 J m2 s K’. Units of all thermaladmittances in J m2 s K1 and all temperatures in Celsius.Park PCI3 PCI3 at Growth/DecayJ.Lc,tymaterial at sunset end of night PCI3LDF 219 0.40 2.7 6.1 3.4OSB 168 0.54 —5.3 —2.7 2.6Plywood 132 0.64 —0.3 1.2 1.5Fir 0 1.00 —0.8 —0.8 0Concrete —371 2.0 —3.1 —15.8 —12.7158may be large, the thermal admittance differential between a park and city could be ofconsiderable importance in creating PCI5s. However, this effect on cooling is establishedsoon after sunset and there is little growth in PCI thereafter.Dry soils and deserts have thermal admittances as low as 600 J m2 s K—’ whilesaturated clay soils and dense paved surfaces may be as high as 2200 J m2 s K—’(Oke, 1987). Therefore, depending on the nature of the urban fabric and the moisturecontent and soil type of the park, differences can be substantial. The SHIM predictionsof the effect of u on UHI development, by Oke et at. (1991) found a heat island of about4°C developed if ururaj/,ucjty = 0.42. This is comparable to the observed NPCI5 whenaUpark/gticity = 0.40 (Table 7.2). Oke et at. (1991) suggested that for typical real worldvalues, the effect of p alone is similar to geometry, and that a 6°C heat island can begenerated if ii rural is very low. Given the results of the scale model here it is possible thatPCI.s larger than 3.4° C can be produced if ,upark is much lower than that of the city.7.3 Evaporative effects on park coolingThe flat, wet model design compares the cooling of both a pan of water, and moist blottingpaper over fir, to the cooling of the background fir. At “sunset” the moist blotting paperis 2°C cooler than the water because the latter absorbed the strong heating better. Atthis time the parks are substantially cooler (by about 24° C) than the surrounding city.After sunset the water “park” with its high heat capacity, cools only by about 3.5°C,while the blotting paper “park”, with lower moisture content, cools by 6°C (Fig. 7.8a).To assess the effect of evaporative cooling as distinct from the thermal admittance effect,a SHIMpDE simulation, initialized with the scale model boundary conditions, was runfor the water. This shows that the effect of evaporative cooling was —1.7° C, about thesame as the effect of thermal admittance.159In this case, evaporative cooling doubles the effect of thermal admittance and makesthe NPCI3 less negative i.e. it increases the temperature differential between the parkand the city (Fig. 7.8b). Due to these low cooling rates the NPCI3 is eroded throughoutthe night . However, the park is sufficiently cooler than the city at sunset that, despitethe slow cooling, it remains a cool island near sunrise.7.4 Combined effects on park cooling7.4.1 Radiative and thermal effectsThe combined effects of surface geometry and thermal admittance on park cooling wereassessed using the model with a dry park and surrounding canyon geometry (canyons,dry park design). Consistent with differences in p, LDF cools by the greatest amountwhile concrete cools the least. The AGEMA images (Fig. 7.9) show the warm concretesurface and the very cool LDF surface with visible view factor effects which agree withFigure 7.1. The plywood and OSB cases are similar at the end of the night but thesurface temperatures of the OSB park are more patchy, due to the spatially—variabledensity of this material.The cooling of points across the different park surfaces is used to illustrate the role ofboth surface geometry and thermal admittance differences (Fig. 7.10). With increasingsky view factor, surface temperatures are expected to decrease. However, in the warmconcrete park, heat is lost by conduction at the edges to the cooler urban surroundings.Hence the reversal in pattern. For the remaining parks with positive differences in it,as /‘ increases from about 0.65 to 0.92, the PCI increases by about 6—8°C. Similarly,an increase in zt of only about 100 J m2 s results in a 3—7°C increase inthe PCI. Radiative cooling appears to have a dominant role. However, the range ofthermal admittance in the model materials (about 150—800 J m2 s K—’) is only 40%160(a) Surface coolingFigure 7.8: Surface cooling (a) and PCIs for parks with evaporating surfaces (b). Atsunset both parks had PCIs of about 24°C. Cooling of the water due to evaporationonly, is shown by E, while cooling due to thermal admittance effects is shown by p.o x xx+xx00000 1++00000000000 00000AA blott.0 water+ (woter)K E (water)AAAC)C00C)C)C)0z0 100 200 300 400 500Time after sunset (s)(b) Decay of PCI—1—2—3—4—5—60—2—4—6—8—10—12—14—16—186000 100 200 300 400 500 600Time after sunset (s)161———:LII—Figure 7.9: Surface temperatures after 625 s of cooling for parks under the canyons, drypark design: (a) LDF; (b) OSB; (c) plywood; and (d) concrete.(a) Low—density fibreboard (LDF) (b) Oriented—strand board (OSB)(c) Plywood (d) Concreteidrizi162of the range of thermal admittances in the real world (about 600—2200 J m2 s K—’).Therefore the role of thermal admittance is underestimated by the scale model. If Lwere proportionately larger, there is the potential for this process to be as important asradiative cooling.121086‘—S. 4C-).-. 20a-2—4—6—8—10 I I I—400 —300 —200 —100 0 100 200 3001u (urban—pork) (J rn-2 si2 K—’)Figure 7.10: The maximum intensity of the PCI in relation to the sky view factor,(shown in brackets) of the park surface, and the difference between the thermal admittance of the city and park environments (z).Results for the combined effects of geometry and thermal admittance compare closelywith the numerical simulations of Oke et al. (1991). Firstly, they suggest that as 1t ruraldecreases the heat island magnitude increases. This is also observed here; a decrease inIpark results in an increase of the PCI. Secondly, they note that with large ,ururaj it is notuncommon for small or negative heat islands to form. In the scale model when upark islarge (e.g. concrete), a small, negative cool island is produced (the park average PCL atthe end of the night is —0.6°C). Thirdly, they note that the most favourable combinationof geometry and thermal properties can cause a heat island of up to 10°C to developafter sunset. For a canyon geometry with H:W of 1.15 and Prural 40% of [tci, (most+ md park (0.92)1/4 way into park (0.88)o edge pork (0.77)o cornerpark(0.65) 4+00 o000163comparable to scale model for LDF), they found maximum heat island developmentof 8.5°C — approximately the same as the scale model estimate of 8.3°C for the PCIgrowth after sunset. Finally, they suggest that the roles of surface geometry and thermaladmittance may have similar importance in urban heat island development. The scalemodel suggests radiative processes are dominant in park cooling, but this is for a limitedrange of urban—park thermal admittance differences. Given the good agreement betweenthe scale model and numerical simulation for this limited range of it is likely that, forlarge Lt, the effect on park cooling may be of a similar magnitude to that by radiativetransfer.7.4.2 Relative contribution of processes to park coolingThree materials (LDF, OSB and plywood) were used in each of the four model designs(flat, dry park; flat, wet park; canyons, dry park; and canyons, wet park). Note that theterm “canyons” means the model has surface geometry with urban canyons and an openpark in the centre. Their behaviour helps elucidate the combined effects of radiative,thermal and evaporative cooling.At sunset the absolute magnitude of the PCI8 is greatest for the canyons, wet parkdesign for all materials (Table 7.3). However, since the model does not try to simulatedaytime park heating, these figures can only be used as a rough guide to indicate whetherthe park is warmer or cooler than the city at sunset.The growth and decay of the PCIs is shown in Figure 7.11. For all three materialsthe dry model scenarios result in growth of the PCI3, particularly if urban canyons arepresent. In contrast, in the case of all wet model parks, the PCI3 is either eroded orthere is no growth. The OSB park, which can hold more moisture, clearly shows thePCI3 is reduced in both wet model cases. Despite this erosion, wet parks still maintainthe highest PCI3s at the end of the night. By 625 s, the LDF park, which has low i,164Table 7.3: Absolute magnitude of the PCI for different model designs. All temperaturesare in Celsius and values are rounded to the nearest integral value.Material and PCI PCI at Growth/Decaymodel design at sunset end of night PCIa) LDFcanyons, dry park 2 10 8flat, dry park 3 6 3canyons, wet park 14 14 0flat, wet park 10 10 0b) OSBcanyons, dry park -4 3 7flat, dry park -5 -3 2canyons, wet park 19 14 —5flat, wet park 11 3 —8c) Plywoodcanyons, dry park 2 8 6flat, dry park 0 2 2canyons, wet park 12 12 0flat, wet park 8 5 -3165almost develops a PCI3 in the canyons, dry park case, comparable to that achieved underthe canyons, wet park design (Table 7.3). This suggests that if a park has low thermaladmittance compared to the city, the magnitude of the PCI3 can approach (or perhapsexceed) that for a moist park.In summary, after sunset and in the absence of geometry, if Ppark << Lcity the PCI3growth is slow and small (about 2°C). The presence of urban canyon geometry (H:W =1) around the park, increases the PCI by about 4—5°C for open grassed parks. Thereis an optimum park width (about 7.5 times the height of the park border) to attainthe maximum amount of radiative cooling. However for the park influence to extendbeyond the border, it is suggested that larger parks have a greater effect. As the treecover increases, the NPCI3 decreases due to reduction of radiative cooling by view—factoreffects. The presence of moisture in the park either prevents further growth beyond thesunset value, or results in a decay of the NPCI3. However, the presence of moisture mayhave created sufficient evaporative cooling by day, that despite this reduced nocturnalcooling, moist parks still have the highest PCI3s by sunrise. Parks with low p and highhave the potential to develop PCI3s comparable in magnitude to those of wet parks.This is due primarily to radiative rather than thermal admittance effects. However, thesmall range of p possessed by the model materials resulted in an underestimation ofthe role of thermal admittance in park cooling. With larger urban—park differences inthermal admittance, it is possible that this process is of equal importance to radiativetransfer. This conclusion is supported by predictions from SHIM, which suggest that theroles of surface geometry and thermal admittance may have similar importance in urbanheat island development. Given the good agreement between the scale model and SHIMresults, this numerical model is recommended as a valuable tool for analysis of the effects166(a) Low—density fibre—board (LDF) (b) Oriented—strand board (ass)10 •I•,•I’II’I 100 flat, dry8 8 + flat, wet0 C yns dry6 64 42—=+z:_________5 0 V g 5 0a2 2—2 0 flat,dry —2 42°<XXXX+ flat wet—4 0 cyns. dry _x cyns, wet ++++XXXXxxxxw—6 —6—8 I I —80 100 200 300 400 500 600 C 100 200 300 400 500 600Time after sunset (s) Time after sunset (s)(c) Plywood10 i... I0 flat, dry8+ flat, weto cyns. dry6cyns, wet4.02—2—4—6—80 100 200 300 400 500 600Time after sunset (s)Figure 7.11: A comparison of PCIs for different materials: (a) LDF; (b) OSB; (c) plywood; under the different model designs (flat, dry; flat, wet; canyons, dry; canyons, wet).In the “canyons” model designs, the open treeless park is surrounded by canyon geometry.167of surface geometry and thermal admittance on park cooling.Thus evaporative cooling is critical in establishing the park as a cool island by sunset.However, after sunset moisture retards park cooling. In the limited boundary conditionsof the model, radiative transfer dominates nocturnal cooling, but in the real world whereurban—park thermal admittance differences are larger, these two processes may be ofequal importance.7.5 Comparison of scale model results with surface cooling observed in Vancouver parksThe scale model results show good agreement with observed nocturnal cooling, at leastin a qualitative sense. The model was not designed to simulate real world differences inmoisture and thermal admittance, but rather to give a first approximation to the effectsof these properties.As indicated by the modelling, both the presence of tree canopies and moisture, arecritical in establishing the park as a cool island by sunset. This is confirmed by daytimeobservations of surface temperatures in Vancouver parks which showed strong temperature contrasts between well—treed and irrigated parks and built surfaces. As a furtherindication of the importance of tree canopies by day, the surveys of air temperatures ineach city generally found large PCI in parks with extensive tree canopies. These parksmay have a PCI that is well established in the afternoon (shade and possibly evaporatively induced). At sunset they have a marked increase in PCI that probably resultsfrom a combination of evaporative and radiative cooling. There are probably minimaleffects of evaporative cooling in lowering air temperature at ground level by day (dueto mixing into the urban boundary layer), but when turbulence and advection decreasenear sunset, this cooler air may remain within the park system. This effect is of short168duration though, and the process of radiative cooling continues to cause the PCI to growthrough the night.Field observations show open parks rapidly cool after sunset and the PCI continues toincrease through the night. In these open parks, the cooling rate may be similar to thatin rural surrounds. However, these parks may be warmer than their urban surrounds byday. Therefore, despite similar rates of cooling between open park and rural surfaces,the nocturnal PCI is less than the UHI.Scale model results also show that park cooling is similar to the rural case. The goodagreement between the scale model results and SHIM simulations indicates that the PCImay in fact be an “inside—out” UHI. However, because open parks may be warmer thantheir urban surrounds at sunset, the PCI is typically less than the UHI.Scale modelling confirms that radiative transfer is an important control on park cooling. The modelling shows that moist parks, despite their slower nocturnal cooling, havethe highest PCI8 by sunrise, due to their coolness at sunset. The model also suggeststhat parks with low p, can attain PCI8 as large as moist parks. Field observations showthat open dry parks cool much more than moist parks. The scale model did not attemptto simulate daytime conditions so the role of moisture at sunset may have been overestimated. The range of j in the model is small but sufficient to indicate likely effects.Even small urban—park differences in thermal admittance can double the park cooling.In the real world, where i differences are large, this effect is accentuated and the coolingof parks with lower surpasses that of moist parks.As a first approximation to the relative contribution of cooling mechanisms, the scalemodel performs satisfactorily. The range of model materials limited the applicability ofresults to real world situations, but they compare well with estimates from numericalsimulations. The model also provides an approximate means to assess the influence ofmoisture on the development of park cool islands.169Chapter 8ConclusionsThis chapter summarizes the major results of the dissertation. This includes conclusionsfrom the survey of the park effect on urban thermal regimes; surface energy balancecomparisons of urban parks in different summer climates; and the relative contributionof processes to nocturnal cooling in urban parks. A discussion of these results leads tothe development of practical guidelines for planners regarding the manipulation of urbanparks to attain maximum climatic benefit.8.1 Summary of results• Surveys of the park influence on the thermal regime of two urban areas with differentsummer climates, both confirms and extends previous studies. The park effect inVancouver which has a cool summer Mediterranean climate, is typically small (1—3°C) but can approach 5°C under ideal conditions. This is higher than previousestimates, but it is certainly not typical. In hot summer Mediterranean climates,the park effect is enhanced and irrigated greenspace can create cool islands of atleast 5—7°C.• Surface temperature contrasts between irrigated greenspace and the built environment, peak in the afternoon but dry grass parks are not necessarily cooler thanurban areas. Shortly after sunset cooling is well established in parks with high skyview factors. By sunrise drier parks, with lower thermal admittance, have cooled170the most. Parks with extensive tree coverage are warmer than open parks at night.• The park effect on air temperatures is smaller by day, except in parks with substantial shade tree borders. The trend of the park cool island (PCI) through thenight suggests that evaporative and radiative cooling processes may be importantin creating a large suburban—park temperature differential soon after sunset. Withreduced turbulence near sunset, in parks with extensive tree canopies, evaporatively cooled air may remain within the park system, lowering air temperatures.This effect is of short duration and the process of radiative cooling then causes thePCI to grow through the night. Open grass parks with higher sky view factors,are dominated by radiative cooling and have a maximum PCI near sunrise. Asthe moisture content (and hence thermal admittance) decrease, the park coolingincreases.• Although air temperatures in a park may be considerably cooler than nearby suburban temperatures, the influence of the park is restricted to the neighbourhoodwithin about one park width.• The surface energy balance of an urban park in a hot, dry summer climate isdominated by evaporation and has consistently negative sensible heat advection inthe late afternoon. Soil heat flux is small. While the park has a similar SEB to thatof a wet rural site, it is distinguished from other land—use types by its propensity toact as an “oasis”. High rates of evaporation are measured in the park, particularlyat its upwind edge. There is an approximately exponential decay of evaporationwith distance into the park. In the late afternoon evaporative forcing is sufficientto induce oasis—type advection.171• The SEB of an unirrigated, but moist, urban park in a temperate climate, is alsodominated by evaporation. Sensible heat is of secondary importance and there islittle enhanced advective. influence on evaporation. Consequently the park doesnot behave as an oasis. It is suggested this is due to the lack of irrigation, because an irrigated suburban lawn in Vancouver does exhibit microscale advectionedge effects. Evaporation across the park varies in response to the turbulence andmoisture regimes.• Comparison of the SEB of urban parks in two cities with different summer climates,highlights the role of soil moisture in determining energy partitioning. In moist orwet urban parks, evaporation dominates the SEB and can exceed the potential rateif surface forcing is sufficient to induce edge— and oasis—type advection. Irrigationis thought to be a criterion for this to occur.• Scale modelling of nocturnal cooling in urban parks provides insight into the relative contribution made by different processes. It suggests that evaporative coolingis critical in establishing the park as a cool island by sunset. However, after sunset,since moisture increases the thermal admittance of the surface, cooling is slowed.With the limited range of model materials, radiation dominated nocturnal cooling with thermal admittance playing a secondary, but important role. However,the modelling suggests that when urban—park thermal admittances are large, thisprocess could be of equal importance to radiative cooling.• Given the good agreement between the results of the scale model and numericalsimulations by the Oke et al. (1991) Surface Heat Island Model (SHIM), thisnumerical model is recommended as a valuable tool for analysis of the effects ofsurface geometry and thermal admittance on park cooling.172• The field research and scale modelling show that urban parks may behave more likerural than urban surfaces. Nocturnal cooling in open parks may be at a rate similarto the rural case. However, because open parks may be warmer than their urbansurrounds during the day, the park cool island is typically less than the urban heatisland at night.8.2 Implications of research for park designAs discussed in the opening chapter, the design of parks incorporates many factors, bothsocial and physical. The ultimate design involves a trade—off between desired characteristics. One of the main objectives of this dissertation is to develop practical guidelinesfor planners regarding the manipulation of urban parks to attain the maximum climaticbenefit. Consideration is only given to the design of parks to cool neighbourhoods. Thisis most applicable to cities in warm climates.Designing a park to achieve the maximum amount of cooling also involves trade—offs.Both daytime and nighttime conditions must be considered. By day the importance oftrees in effecting cooling at ground level, has been confirmed. They lower air temperaturesat ground level mainly through the provision of shade. Evaporative cooling, is likely tohave a negligible impact on air temperatures at ground level as the cool air is rapidlymixed into the urban boundary layer. On the other hand, at night cooling is reducedunder a tree canopy because the sky view factor is reduced. Yet this is the critical timewhen cooling of a few degrees can bring relief to neighbourhood communities. Thereforethere must be a trade—off to find an appropriate design that can offer relief from hightemperatures in both the day and nighttime.Observations in parks, together with the scale modelling results, indicate the importance of evaporative cooling in establishing parks as cool islands by sunset. Parks with173substantial tree canopies and well irrigated greenspace show an increase in the PCI soonafter sunset. It is suggested that this is partly due to evaporative cooling. This effectlasts only a few hours and unfortunately the presence of moisture (through increasingthermal admittance) and trees (through decreasing the sky view factor), slows cooling.Therefore the optimum design might be a savannah—type park with loose clusters oftrees interspersed by wide open, irrigated grass. The irrigated greenspace contributes toan increase in the PCI at sunset and the open spaces promote rapid radiative cooling toincrease the PCI through the night.The arrangement of trees must be carefully planned. Dense plantings around the parkborder should be avoided because these impede air movement both into, and out of, thepark. Rather, clusters of trees should be interspersed around the edges. Careful attentionshould be given to the prevailing wind regime on summer evenings. There should be fewtrees at the prevailing downwind edge of the park to allow advection of cooler air intothe neighbourhood.These guidelines suggest that multi—use parks and golf courses with combinations oftrees and open spaces, are among the best park designs. However, golf courses often havedense vegetative borders that restrict airflow into neighbourhoods. Field observations ofthe park effect in this study, confirm that these parks achieve some of the highest coolingrates of park types and attain maximum cooling soon after sunset, i.e. at the time whenlower temperatures are particularly beneficial for neighbouring communities.The optimum park size depends mainly on the geometry of the urban surrounds. Toattain a significant amount of radiative cooling, the park width should be at least 7.5times the height of the park border. The park effect, however, remains fairly localizedinfluencing only the adjacent neighbourhood up to about one park width away. Howeveras the park size increases, a larger volume of air is cooled which increases the potentialfor advection of cool air beyond the park boundaries. This suggests the need for many174interspersed neighbourhood parks to attain the maximum cooling benefit.The central objective of this dissertation is to increase understanding of the energeticsand cooling in urban parks. An integrated research approach has contributed to achievingthis objective. Knowledge of the park effect on urban temperatures has been increasedthrough detailed spatial and temporal surveys in cities with different climates. Thecauses of this park cooling have been elucidated through scale modelling which confirmsand extends, existing numerical model results. This dissertation also presents the firstmeasurements of the surface energy balance of urban parks. Measurements in parks intwo cities with different summer climate, has improved understanding of park energetics.This field component assessed both the spatial variability of the SEB in the parks, as wellas placing the park in context with comparisons to energy partitioning in nearby land—usetypes. Finally, this dissertation provides some practical guidelines to planners suggestinghow urban parks should be designed to achieve maximum cooling. This important linkbetween process studies and practical applications, is often neglected.175ReferencesAkbari, H., Huang, J., Martien, P., Rainier, L., Rosenfeld, A., and Taha, H. 1988: Theimpact of summer heat islands on cooling energy consumption and CO2 emissions.In Proceed. 1988 Summer Study in Energy Efficiency in Buildings. Amer. Counc.for an Energy—efficient Economy, Washington, D.C.Akbari, H. and Taha, H. 1992: The impact of trees and white surfaces on residentialheating and cooling energy use in four Canadian cities. Energy. 17: 141—149.Baldocchi, D.D. 1989: Canopy—atmosphere water vapour exchange: can we scale froma leaf to a canopy? In Estimation of Areal Evapotranspiration. Black, T.A., Spittlehouse, D.L., Novak, M.D. and Price, D.T. (eds.). I.A.H.S. Publ. No. 177.p.21—41.Bergen, J.D. 1975: Air movement in a forest clearing as indicated by smoke drift. Agric.Meteorol. 15: 165—179.Bernatzky, A. 1982: The contribution of trees and green spaces to a town climate.Energy and Buildings 5: 1—10.Blad, B.L. and Rosenberg, N.J. 1974: Evapotranspiration by subirrigated alfalfa andpasture in the east central Great Plains. Agron. J. 66: 248—252.Brakke, T.W., Verma, S.B. and Rosenberg, N.J. 1978: Local and regional componentsof sensible heat advection. J. Appi. Meteorol. 17: 955—963.176Brest, C.L. 1987: Seasonal albedo of an urban/rural landscape from satellite observations. J. Gum. Appi. Meteorol. 26: 1169—1187.Brunt, D. 1941: Physical and Dynamical Meteorology. Cambridge University Press:London. 428p.Brutsaert, W.ll. 1982: Evaporation into the Atmosphere: Theory, History, and Applications. D. Reidel Publishing Co.: Dordrecht. 299p.Chandler, T.J. 1965: The Climate of London. Hutchinson: London.Chapman, A.J. 1984: Heat Transfer. MacMillan: New York. 6O8p.Chen, J.M. and Black, T.A., Novak, M.D., and Adams, R.S. 1994: A wind tunnel study’of turbulent air flow in forest clearcuts. Agric. Forest Metcorol. (In press.)Chen, J.M., Ibbetson, A. and Milford, T.R. 1988a: Boundary—layer resistances of artificial leaves in turbulent air:I Leaves parallel to the mean flow. Bound.—LayerMeteorol. 45: 137—156.Chen, J.M., Ibbetson, A. and Milford, T.R. 1988b: Boundary—layer resistances of artificial leaves in turbulent air:II Leaves inclined to the mean flow. Bound.—LayerMeteorol. 45: 371—390.Clark, G.A., Reddell, D.L., Adams, R.F. 1984: Evaluation of mini—lysimeters for deficitirrigation studies. ASAE Technical Paper, 84—262 1.Clarke, J.F. and Bach, W. 1971: Comparison of comfort conditions in different urbanand suburban microenvironments. Jut. Jour. Biometeorol. 15: 41—57.177Cleugh, H.A. 1988: Development and evaluation of a suburban evaporation model: astudy of surface and atmospheric controls on the suburban evaporation regime.Unpubi. Ph.D. Thesis. The University of British Columbia, Vancouver.Cleugh, H.A. and Oke, T.R. 1986: Suburban—rural energy balance comparisons in summer for Vancouver, B.C. Bound.—Layer Meteorol. 36: 351—369.Coppin, P.A., Raupach, M.R. and Legg, B.J. 1986: Experiments on scalar dispersionwithin a model plant canopy. Part II: An elevated plane source. Bound.—LayerMeteorol. 35: 167—191.Davies, J.A., Robinson, P.J. and Nunez, M. 1971: Field determination of surface emissivity and temperature for Lake Ontario. J. Appi. Meteorol. 10: 811—819.Dugas, W.A. and Bland, W.L. 1989: The accuracy of evaporation measurements fromsmall lysimeters. Agric. Forest. Meteorol. 46: 119—129.Dugas, W.A., Fritschen, L.J., Gay, L.W., Held, A.A., Matthias, A.D., Reicosky, D.C.,Steduto, P. and Steiner, J.L. 1991: Bowen ratio, eddy correlation, and portablechamber measurements of sensible and latent heat flux over irrigated spring wheat.Agric. Forest Meteorol. 56: 1—20.Dwyer, J.F., McPherson, E.G., Schroeder, H.W. and Rowntree, R.A. 1992: Assessingthe benefits and costs of the urban forest. J. Arboriculture. 18: 227—234.Dyer, A.J. and Crawford, T.V. 1965: Observations of the microclimate at a leadingedge. Quart. J. Roy. Meteorol. Soc. 91: 345—348.Eliasson, I. 1993: Urban climate related to street geometry. Unpubi. Ph.D. Thesis,Göteburg University, Sweden.178Fairey, P. and Kalaghchy, S. 1982: Evaluation of thermocouple installation and mounting techniques for surface temperature measurement in dynamic environments. InSeventh Nat. Passive Solar Conf. Hayes, J. and Winn, C.B. (eds.), Am. SolarEnergy Soc. Inc.Fritschen, L.J. and Gay, L.W. 1979: Environmental Instrumentation. Springer—Verlag:New York.Fritschen, L.J. and van Bavel, C.H.M. 1962: Energy balance components of evaporatingsurfaces in arid lands. J. Geophys. Res. 67: 5179—5185.Gash, J.H.C. 1986: Observations of turbulence downwind of a forest—heath interface.Bound.—Layer Meteorol. 36: 227—237.Givoni, B. 1991: Impact of planted areas on urban environmental quality: a review.Atmos. Environ. 25B: 289—299.Gold, E. 1956: Smog. The rate of influx of surrounding cleaner air. Weather 11:230—232.Goltz, S.M. and Pruitt, W.O. 1970: Spatial and temporal variations of evapotranspiration downwind from the leading edge of a dry fallow field. Div. Tech. rep.ECOM68—G1O—1. Dept. of Water Sci. and Engineering, Univ. of California—Davis.Grey, G.W. and Deneke, F.J. 1978: Urban Forestry. Jhn Wiley: New York.Grimmond, C.S.B. 1991: An objective urban heat storage model and its comparisonwith other schemes. Atmos. Environ. 25 B: 311—326.Grimmond, C.S.B. 1988: An evapotranspiration—interception model for urban areas.Unpubl. Ph.D. Thesis. The University of British Columbia, Vancouver.179Grimmond, C.S.B., Isard, S.A. and Belding, M.J. 1992: Development and evaluation ofcontinuously weighing mini—lysimeters. Agric. and Forest Meteorol. 62: 205—218.Grimmond, C.S.B. and Oke, T.R. 1994: Comparison of heat fluxes from summertimeobservations in the suburbs of four North American cities. Submitted to J. Appi.Meteorol.Grimmond, C.S.B., Oke, T.R., and Cleugh, H.A. 1993: The role of “rural” in comparisons of observed suburban—rural flux differences. Exchange Processes at the LandSurface for a Range of Space and Time Scales. Proceed. Yokohama Symposium,July, 1993. JARS Publ. No. 212. p.165—74Heisler, G.M. 1986: Effects of individual trees on the solar radiation climate of smallbuildings. Urban Ecology 9: 337—359.Heisler, G.M. 1982: Reductions of solar radiation by tree crowns. In The RenewableChallenge, Proc. Ann. Meeting Amer. Sector of ml. Solar Energy Soc. Houston,Texas. Amer. Solar Energy Soc., Boulder, CO. p.133—139.Heisler, G.M. 1977: Trees modify the metropolitan climate and noise. Jour. Arboric.3: 201—207.Herrington, L.P. 1977: The role of urban forests in reducing urban energy consumption. In Forests for People, Proc., Society of Amer. Foresters. Washington, D.C.p.60—66.Herrington, L.P. 1976: Vegetation and thermal environments of human settlements. InTrees and forests for human settlements. P 1.05 Project Group on Aboriculture andUrban Forestry.180Herrington, L.P., Bertolini, G.E. and Leonard, R.E. 1972: Microclimate of a suburbanpark. In Proc. Conf. on the Urban Environ. A.M.S, Boston. p.43—44.Honjo, T. and Takakura, T. 1991: Simulation of thermal effects of urban green areason their surrounding areas. Energy and Buildings. 15—16: 443—446.Höppe, P. 1991: Improving indoor thermal comfort by changing outdoor conditions.Energy and Buildings 15—16: 743—747.Hosker, R.P. Jr. 1986: Flow around isolated structures and building clusters: a review.ASHRAE Trans. 1985 91(2B): 1671—1692.Hosker, R.P. Jr. 1984: Flow dispersion near obstacles. In Atmospheric Science andPower Production. Randerson, D. (ed.) U.S. Dept. of Energy/Technology Info.Centre D OE/TIC—27601. Springfield, VA.Howell, J.R. 1982: A Catalog of Radiation Configuration Factors. McGraw—Hill: NewYork.Huang, Y.J., Akbari, H., Taha, H., Rosenfeld, A.H. 1987: The potential of vegetation inreducing summer cooling loads in residential buildings. J. Clim. Appi. Meteorol.26: 1103—1116.Hunt, J.C.R., Abell, C.J., Peterka, J.A. and Woo, H. 1978: Kinematical studies ofthe flows around free or surface mounted obstacles: applying topology to flowvisualization. J. Fluid Mech. 86: 179—200.Hutchinson, B.A., Taylor, F.G. and the Critical Review Panel, 1983: Energy conservation mechanisms and potentials of landscape design to ameliorate building microclimates. Landscape J. 2: 19—39.181Isard, S.A. and Belding, M.J. 1989: Evapotranspiration from the alpine tundra of Colorado, U.S.A. Arctic and Alpine Res. 21: 71—82.Jarvis, P.C. and McNaughton. K.G. 1986: Stomatal control of transpiration: scalingup from leaf to region. Adv. in Ecol. Res. 15: 1—49.Jauregui, E. 1991: Influence of a large urban park on temperature and convective precipitation in a tropical city. Energy and Buildings. 15—16: 457—463.Jauregui, E. 1973: Untersuchungen urn Stadtklima von Mexico—Stadt. Unpubi. Ph.DThesis. Rheinischen Friedrich—Wilhelms—Universitãt.Johnson, G.T. and Watson, I.D. 1988: Determining heat transfer parameters in urbanenvironments using non—linear optimisation techniques. Proceed. of the 1987 mt.Conf. on Computational Techniques and Applications. Sydney, Australia, North—Holland: Amsterdam.Johnson, G.T. and Watson, I.D. 1987: Modelling the radiative heating and cooling ofurban canyons. In Proc. mt. Conf on Modelling and Simulation. Melbourne,Australia.Johnson, G.T., Oke, T.R., Lyons, T.J., Steyn, D.C., Watson, I.D. and Voogt, J.A. 1991:Simulation of surface urban heat islands under ‘ideal’ conditions at night. Part 1:Theory and tests against field data. Bound.—Layer Meteorol. 56: 275—294.Kirby, S.F. and Sellers, W.D. 1987: Cold air drainage and urban heating in Tucson,Arizona. J. Arizona—Nevada Acad. of Science. 22: 123—128.Kondo, M and Suzuki, S. 1983: A quantitative study on an existence effect of the urbanvegetation by measuring the surface temperature. J. Japanese Inst. LandscapeArch. 47: 1—15.182Kristensen, L. and Fitzjarrald, D. 1984: The effect of line averaging on scalar fluxmeasurements with a sonic anemometer near the surface. J. Atmos. Oceanic Tech.1: 138—146.Landsberg, H.E. 1981: The Urban Climate. Academic Press: New York, 275 pp.Lang, A.R.G., Evans, G.N. and Ho, P.Y. 1974: The influence of local advection onevapotranspiration from irrigated rice in a semi—arid region. Agric. Meteorol. 13:5—13.Latimer, J.R. 1972: Radiation measurement. Tnt. Field Year for the Great Lakes. Tech.Manual Serv. 2. NRC/USNAS/IHD, Ottawa.Leuning, R., Denmead, O.T., Lang, A.R. and Ohtaki, E. 1982: Effects of heat and watervapour transport on eddy covariance measurement of CO2 fluxes. Bound.—LayerMeteorol. 23: 209—222.Lindqvist, S. 1992: Local climatological modelling for road stretches and urban areas.Geograf. Annaler. 74A: 265—274.Lowe, P.R. 1976: An approximating polynomial for the computation of saturationvapour pressure. J. Appi. Meteorol. 16: 100—103.Lowry, W.P. 1988: Atmospheric ecology for designers and planners. Peavine Publications: McMinnville, Oregon.McNaughton, K.G. 1989: Micrometeorology of shelter belts and forest edges. Phil.Trans. R. Soc. Lond. B 324: 351—368.183McNaughton, K.G. 1988: Effects of windbreaks on turbulent transport and microclimate. In Proc. mt. Symp. Windbreak Technology Brandle, J.R. and Hintz, D.L.(eds.) Ecosys. Environ. 22/23: 17—39.McNaughton, K.G. 1983: The direct effect of shelter on evaporation rates: Theory andan experimental test. Agric. Meteorol 29: 125—136.McNaughton, K.G. and Jarvis, P.G. 1983: Predicting the effects of vegetation changeson transpiration and evaporation. In Water Deficits and Plant Growth, Vol 7Kozlowski, T.T. (ed.) Academic Press: New York.McNaughton, K.G. and Spriggs, T.W. 1989: An evaluation of the Priestley and Taybr equation and the complementary relationship using results from a mixed—layermodel of the convective boundary layer. In Estimation of A real Evapotranspiration.Black, T.A., Spittlehouse, D.L., Novak, M.D. and Price, D.T. (eds.). I.A.H.S. Pubi.No. 177. p.89—104McPherson, E.G. 1989: Effects of three landscape treatments on residential energy andwater use in Tucson, Arizona. Energy and Buildings. 13: 127—138.Monteith, J.L. 1963: Gas exchange in plant communities. In Environmental Control ofPlant Growth. Evans. L.T. (ed.) Academic Press: New York.Myrup, L.O. 1969: A numerical model of the urban heat island. J. Appi. Meteorol. 8:908—918.Nakamura, Y., and Oke, T.R. 1988: Wind, temperature and stability conditions in aneast—west oriented urban canyon. Atmos. Environ. 22: 2691—2700.Oke, T.R. 1989: The micrometeorology of the urban forest. Phil. Trans. R. Soc. Lond.B 324: 335—349.184Oke, T.R. 1988: The urban energy balance. Frog. Phys. Geogr. 12: 471—508.Oke, T.R. 1987: Boundary Layer Climates. Methuen: New York.Oke, T.R. 1984: Methods in urban climatology. In Applied climatology (ed. W. Kirchhofer, A. Ohmura and H. Wanner) (Zurcher Geog. Schriften 14). Zurich: Eidgenossische Technische Hochschule Geographische Institut. p.19—29.Oke, T.R. 1982: The energetic basis of the urban heat island. Quart. J. Roy. Soc. 108:1—24.Oke, T.R. 1981: Canyon geometry and the nocturnal urban heat island: comparison ofscale model and field observations. J. Climatol. 1: 236—254.Oke, T.R. 1979: Advectively—assisted evapotranspiration from irrigated urban vegetation. Bound.—Layer Meteorol. 17: 167—173.Oke, T.R., Cleugh, H.A., Grimmond, C.S.B., Schmid, H.P. and Roth, M. 1989: Evaluation of spatially—averaged fluxes of heat, mass and momentum in the urbanboundary layer. Weather and Climate. 9: 14—22.Oke, T.R. and East, C. 1971: The urban boundary layer in Montréal. Bound.—LayerMeteorol. 1: 411—437.Oke, T.R. and Hay, J. 1994: The Climate of Vancouver. B.C. Geographical Series, No.50. Tantalus Res. Ltd: Vancouver.Oke, T.R., Johnson, G.T., Steyn, D.C. and Watson, I.D. 1991: Simulation of surfaceurban heat islands under ‘ideal’ conditions at night. Part 2: Diagnosis of causation.Bound.—Layer Meteorol. 56: 339—359.185Oke, T.R. and McCaughey, J.H. 1983: Suburban—rural energy balance comparisons forVancouver, B.C. An extreme case? Bound.—Layer Meteorol. 26: 337—354.O’Rourke, P.A. and Terjung, W.H. 1981: Urban parks, energy budgets and surfacetemperatures. Arch. Meteorol. Geoph. Bioki. Ser. B 29: 327—344.Parker, J.H.. 1989: The impact of vegetation on air conditioning consumption. Controlling summer heat islands, Proc. Workshop on Saving Energy and ReducingSummer Heat Islands. Akbari, H., Garbesi, K. and Martien, P. (eds.) Universityof California, Berkeley, California, 45—52.Parker, J.H. 1987: The use of shrubs in energy conservation planning. Landscape J. 6:132—139.Parker, J.H. 1983: The effectiveness of vegetation on residential cooling. Passive SolarJ. 2: 123—132.Parker, J.H. 1981: Landscaping to reduce the energy used in cooling buildings. J.Forestry. 81: 82—84.Priestley, C.H.B. and Taylor, R.J. 1972: On the assessment of surface heat flux andevaporation using large—scale parameters. Mon. Weather Rev. 100: 81—92.Raine, J.K. and Stevenson, D.C. 1977: Wind protection by model fences in a simulatedatmospheric boundary layer. J. Indust. Aerodyn. 2: 159—180.Rauner, J.L. and Cernavskaja, M.M. 1972: The heat balance of a city and the effectof urban green areas on the temperature regime. Akademiya Nauk SSSR, Moscow,Izvestiya, Ser. Geograficheskaya. 5: 46—53.186Raynor, G.S. 1971: Wind and temperature structure in a coniferous forest and a contiguous field. Forest Sci. 3: 351—363.Rider, N.E., Philip, J.R. and Bradley, E.F. 1963: The horizontal transport of heatand moisture — a micrometeorological study. Quart. J. Roy. Meteorol. Soc. 89:507—631.Rijks, D.A. 1971: Water use by irrigated cotton in Sudan: III. Bowen ratios and advective energy. J. Appi. Ecol. 8: 643—663.Robins, A.G. and Castro, I.P. 1977: A wind tunnel investigation of plume dispersionin the vicinity of a surface mounted cube. I. The flow field. Atmos. Environ. 11:29 1—297.Rosenberg, N.J. 1972: Frequency of potential evapotranspiration rates in’ the centralGreat Plains. J. Irrig. Drainage Div. ASCE. 98: 203—206.Rosenberg, N.J. 1969: Advective contribution of energy utilized in evapotranspirationby alfalfa in the east central Great Plains. Agric. Meteorol. 6: 179—184.Rosenberg, N.J., Blad, B.L. and Verma, S.B. 1983: Microclimate — the Biological Environment. John Wiley & Sons: New York.Rosenberg, N.J. and Verma, S.B. 1978: Extreme evapotranspiration by irrigated alfalfa:A consequence of the 1976 midwestern drought. J. Appl. Meteorol. 17: 934—941.Ross, S.L. and Oke, T.R. 1988: Tests of three urban energy balance models. Bound.—Layer Meteorol. 44: 73—96.Roth, M. 1991: Turbulent transfer characteristics over a suburban surface. Unpubi.Ph.D. Thesis. The University of British Columbia, Vancouver.187Roth, M. and Oke, T.R. 1994: Relative efficiencies of turbulent transfer of heat, massand momentum over a patchy urban surface. Submitted to J. Atmos. Sci.Ruck, B. 1993: Wind—tunnel measurements of flow field characteristics around a heatedmodel building. J. Wind Eng. Indust. Aerodyn. 50: 139—152.Schmid, H.P. 1988: Spatial scales of sensible heat flux variability. Representativenessof flux measurements and surface layer structure over suburban terrain. Unpubl.Ph.D. Thesis. The Uiversity of British Columbia, Vancouver.Schmid, H.P., Cleugh, H.A., Grimmond, C.S.B. and Oke, T.R. 1991: Spatial variabilityof energy fluxes in suburban terrain. Bound.—Layer Meteorol. 54: 249—276.Schuepp, P.R. 1973: Model experiments on free—convection heat and mass transfer ofleaves and plant elements. Bound.—Layer Meteorol. 3: 454—467.Schuepp. P.11. 1972: Studies of forced—convection heat and mass transfer of flutteringrealistic leaf models. Bound.—Layer Meteorol. 2: 263—274.Sham, S. 1991: Urban climatology in Malaysia: an overview. Energy and Buildings15—16: 105—117.Sham, S. 1987: Climate amelioration of an urban forest: a study of the Bukit Nenas reserve, Kuala Lumpur. In Tropical Urban Ecosystems Studies, Vol 3: Meteorologicaland Air Quality Impact of the Kiang Valley Region. Sham, S. (ed.) p.111—iTT.Sham, S. 1986: Moderating urban temperatures through tree planting in the KualaLumpur— Petaling Jaya Area, Malaysia. In A Study of the Urban Ecosystem of theKelang Valley Region, Malaysia, Vol. 1: The Kelang Valley Region: Some SelectedIssues. Sham, S. (ed.) p.135—iS2188Siegner, I., Muihearn, P.J., Bradley, E.F. and Finnigan, J.J. 1976: Turbulent flow in amodel plant canopy. Bound.—Layer Meteorol. 10: 423—453.Slatyer, R.O. and Mcllroy, I.C. 1961: Practical Microclimatology with Special Referenceto the Water Factor in Soil—plant—atmosphere Relationships. Unesco, CSIRO.Souch, C.A. and Souch, C. 1993: The effect of trees on summertime below canopy urbanclimates: a case study Bloomington, Indiana. J. Arboriculture 19: 303—311.Steyn, D.C. 1980: The calculation of view factors from fisheye—lens photographs. Atmos. Ocean. 18: 254—258.Steyn, D.C. and Lyons, T.J. 1985: Comments on the determination of view factors inurban canyons. J. Clim. Appi. Meteorol. 24: 383—385.Sundersingh, S.D. 1991: Effect of heat islands over urban Madras and measures for itsmitigation. Energy and Buildings 15—16: 245—252.Taha, H.G., Akbari, H. and Rosenfeld, A.H. 1988: Vegetation canopy micro—climate: afield project in Davis, California. Lawrence Berkeley Lab. Rep. 24593. Berkeley,CA.Tanner, C.B. 1957: Factors affecting evaporation from plants and soils. J. Soil WaterConserv. 12: 221—227.Tanner, B.D. and Greene, J.P. 1989: Measurements of sensible heat and water vapourfluxes using eddy correlation methods. Final Report. Prepared for U.S. ArmyDugway Proving Grounds, Dugway, Utah.van Eimern, J., Karschan, R., Razumova, L.A., and Robertson, G.W. 1964: Windbreaksand Shelterbelts. WMO Tech. Note No. 59. World Meteorol. Organiz.: Geneva.189Vukovich, F.M. 1983: An analysis of the ground temperatures and reflectivity patternabout St. Louis, Missouri, using HCMM satellite data. J. Gum. Appl. Meteorol.22: 560—571.Webb, E.K., Pearman, G.I. and Leuning, R. 1980: Correction of flux measurements fordensity effects due to heat and water vapour transfer. Quart. J. Roy. Meteorol.Soc. 106: 85—100.Weiss, A. and Lukens, D.L. 1981: An electronic circuit for the detection of leaf wetnessand a comparison of two sensors under field conditions. Plant Dis. 65: 41—43.White, J.M., Eaton, F.D. and Auer, A.H. Jr. 1978: The net radiation budget of the St.Louis metropolitan area. J. Appi. Meteorol. 17: 593—599.Whiten, A.J. 1956: The ventilation of Oxford Circus. Weather 11: 227—229.Woo, H.G.C., Peterka, J.A. and Cermak, J.E. 1977: Wind—tunnel measurements in thewakes of structures. NASA Contractor Rep. NASA CR—2806. Springfield, VA.Wright, J.L. and Jensen, M.E. 1972: Peak water requirements of crops in southernIdaho. J. Irrig. Drainage Div. Proc. ASCE. 98: 193—201.190Appendix A: Survey of the Park Effect in Tucson, ArizonaThis appendix presents the results of a pilot survey of the park influence on air temperatures in Tucson, Arizona. Tucson has a very hot dry summer climate (Köppen climateclassification BWh — arid desert). The Tucson basin (32° 07’N, 1100 56’W) lies at an elevation of about 730 m and is bounded on three sides by the Santa Catalina, Tucson andRincon Mountains to the north, west and east, respectively. The Santa Rita Mountainslie about 100 km to the south. The topography is flat or gently rolling with many drywashes. Tucson is located in a desert environment predominantly vegetated by creosotebushes (Larrea divaricata). The population of metropolitan Tucson is approximately700,000.Eight traverses measuring air temperature were conducted in May/June, 1990. Thesynoptic conditions throughout the observation period were controlled by the presence ofa thermal low over Arizona. Tucson has a large number of urban parks. The vegetativecomposition of the parks depends on the amount of irrigation applied; several are naturaldesert, but most are irrigated. The eighteen parks surveyed range in size from 2 to 140 ha(Fig. A.1). Most of the parks have 50—80% of their surface area in turf, some are fullyirrigated and one is a large (111 ha) natural desert park. The majority are multi—useparks with savannah parkland and playing fields — particularly baseball diamonds, whichare intensely used especially in the evenings. Four golf courses adjacent to the largerparks, were also surveyed.Eight traverses around Tucson parks were made in May/June 1990 under mainlyclear skies, and a range of wind conditions. Because of breezy conditions only three ofthe more extensive traverses (through Tucson parks to the desert) were conducted. The191Figure A.1: Location of the fixed sites and the traverse routes for the park surveys inTucson.KeytoTuscon Parks1. Reid Park2. Randolph GolfCourse3. Freedom Pork4. Lincoln Park5. Santa Rita Park6. De Anza Park192remaining traverses concentrated only on Reid Park. A summary of the PCI and UHIis given in Table A.l. Overall PCI are relatively large. The maximum PCI is 6.8°C.This occurs with high scattered cloud when average wind speed was lowest (1.5 mReid Park (see Fig. A.1), a large (53 ha) multi—use park, consistently exhibits the highestPCI. Lincoln Park, which jointly shares the greatest cool island intensity with Reid Parkon June 2, is a similar large (77 ha) multi—use park. The smallest PCI (1.2°C) occurswhen the average wind speed is the largest (3.7 ms1). This combination of cloud coverand wind suggests that radiation may be less critical than turbulence as a control onthe magnitude of the temperature differential. The urban heat island intensity is low,ranging from 1.4°C to 4.6°C.Table A.1: Summary of the nocturnal thermal climate of parks in Tucson. City—widetraverses only occurred on May 15, May 24 and June 2. The remaining traverses onlysurveyed Reid Park. All traverses began an hour after sunset.Date Park Maximum Range UHI Mean wind Cloud1990 PCI (°C) PCI (°C) (°C) speed (m s’) coverMay 15 Reid 3.3 0.2—3.3 1.4 2.6 high sctMay 19 Reid only 1.2 3.7 clearMay 20 Reid only 2.7 2.1 clearMay 22 Reid only 6.8 1.5 high sctMay 23 Reid only 1.5 3.4 mid brknMay 24 Reid 3.0 0.2—3.0 4.6 2.7 clearJune 1 Reid only 3.3 3.2 clearJune 2 Reid, Lincoln 1.9 0.7-1.9 2.6 2.8 clearTemperature traverses on calm, clear nights, show higher temperatures in the commercial and downtown regions (Fig. A.2). The warmest area was a commercial routenortheast of the city centre. In the suburbs temperatures steadily decline towards the193surrounding desert except for even cooler spots in urban parks.Observations show that concentrated human activity, such as a large baseball game,can offset park cooling. The largest PCI was observed on nights without such activity.The Reid Park traverse on May 22 (Fig. A.3) was repeated three times (in consecutiveruns) to verify the almost startling results. A large temperature difference of nearly 7°Cwas observed between the nearby commercial strip along Speedway Boulevard, and thecentre of the greenspace consisting of Reid Park and Randolph Golf Course. The lightwesterly winds can be seen to advect cooler air to the east of the park. The warmer areato the north and southeast, are associated with commercial strips.The magnitude and extent of nocturnal cooling in urban parks in Tucson exceeds thatcommonly observed in mid—latitude temperate cities. On the other hand, the magnitudeof the urban heat island shortly after sunset is relatively small compared to that of mid—latitude cities of comparable population in similar conditions. Continuous measurementsof air temperature at a suburban (Treat) and a desert site (Houghton) (Fig. A.l) for 20days in May/June 1990, showed that Tucson’s UHI was small (< 1°C) and often negativeby day, and peaked at 4—5° C sometime between midnight and sunrise (Grimmond andOke, pers. comm.).194C-)0G)CQ)ELegend:1. Reid Park (multi-use)2. Randolph Golf Course3. Freedom Park (multi-use)4. Lincoln Park (multi-use)5. Santa Rita Park (multi-use)6. De Anza Park (multi-use)29282726252423Figure A.2: Air temperatures for the downtown to desert traverse an hour after sunset,May 24, 1991.o 10 20 30 40 50 60 70 80 90 100 110Time elapsed (mm)195Figure A.3: Isotherm map for Reid Park, May 22. Arrows indicate the direction oftraverse and the spots are check—points for the traverse.Legendc Isotherm• CheckpointN‘ Trgverse AScale:6196Appendix B: Errors in Scale ModellingB.1 IntroductionThe probable error analysis presented by Fritschen and Gay (1979) is used as a frameworkto estimate errors in measured variables for the scale modelling. The absolute errorof a function Y consisting of variables x1,x2, ...x1, is estimated:ix18Y tx2OY i2xôY+ + (B.1)where are the absolute errors of each of the variables. This provides a “worst case”estimate of the errors assuming that all the errors act in the same direction. It is morelikely that the errors about x1 are normally distributed and there is some probability thaterrors in different variables may offset each other to a limited degree. The probable errorin Y, Y, can be estimated by combining the individual errors through a least squaresapproach:.(B.2)\ ôxlj 8X2The differentials of Y with respect to each of the x are substituted into (B.2) andtypical values of the variables are used to generate W. The values of Sx1 are the errorsin the variables, typically the root mean square error (RMSE), and are obtained fromcomponent instrument errors, or estimated by other means.B.2 Surface temperature errorsIn the scale modelling there are several potential errors associated with measurement ofsurface temperature by thermocouples. These include errors due to the thermocouple197wire, the use of the 21X micrologger to measure the voltage and linearize the data, andmounting errors (Table B.1).Table B.1: Summary of errors in surface temperature measurement by thermocouples.Type of error Amount of % of totalerror (°C) error (°C)Reference temperature 0.8 61Thermocouple output 0.5 24Voltage measurement 0.07 < 1Reference linearization 0.001 < 1Output linearization 0.001 < 1Mounting 0.4 15Sum 1.05RMSE 1.03The main sources of error are the reference temperature, thermocouple output, an&mounting errors. The reference temperature error results from a combination of errors inthe reference thermistor (± 0.5°C in the range 20—70°C) and when there is a differencebetween the thermistor and the actual reference junction (the channel to which thethermocouple is attached). This latter error has a maximum of 0.3°C in field situations(CSI manual, 1991) but is likely to be less in a laboratory situation.The thermocouple output error is given by the manufacturers standard for type—Tthermocouple wire. The mounting error includes radiation and conduction errors whichresult from the attachment of the thermocouple to the model surface. An estimate ofthe error is made using the range of differences between temperature measured by thethermocouples and by the AGEMA.The voltage measurement error is generally 0.05% of the full scale range used to198measure thermcouple voltage. On the 5 mV range used, this results in a temperatureerror of 0.06—0.07°C for the range of temperatures encountered. The reference and output linearization errors result from approximations used in the conversion of voltage totemperature and are negligible.The combined errors give a RMSE of 1.03°C in surface temperature measurement bythe thermocouples.B.3 Emissivity errorsThe emissivity () of the model surface was estimated by the method of Davies et al.(1971):(B.3)where T is the apparent surface temperature; T is the true surface temperature (measured while covered with a polished aluminium cone) and Tk is the apparent radiativesky temperature. These temperatures were measured with an Everest 4000A infra—redthermometer and logged onto a CR21X data—logger. The associated errors are 0.5°C forthe infra—red thermometer and 0.03° C for the voltage measurement on the data—logger.This gives a RMSE of 0.50° C.The probable error in emissivity estimates is given by:[QTr)2+(6T-)2+(STk)2]r (B.4)with partial errors:4T, B55T(T—T,),1rr3frr4 p4Vf sLr Ak B65T3 (T—T,92AP3I4 p4‘k11r ‘k) 1B7(T34— ,92199Typical values of Tr, T3 and Tk (285.67, 286.87 and 244.18 K respectively) weresubstituted into equation B.4 giving a probable error of 0.020 in the surface emissivitywhen the errors of Tr, 7’ and Tk are set to 0.5 K.200Appendix C: Sky View Factor CalculationsThis appendix outlines three methods used to estimate sky view factors (fr) for locationsin the scale model. The first, derived from Steyn and Lyons (1985), is used to estimate bfor canyon and park geometries in the absence of vegetation. For simple park geometries(e.g. grass parks with “deciduous tree” borders), a method from Howell (1982) is usedto estimate ,b5. Finally, for more complex arrangements of vegetation (savannah, gardenand forest parks), is calculated from fisheye—lens photography using the method ofSteyn (1980).C.1 Calculation of for locations in grass parks and canyons.In the simplest case with no vegetation, V5s is calculated from an analysis derived fromSteyn and Lyons (1985) and Steyn, (pers. comm.):(C.1)where is the view factor for a wall, calculated from:(x,y) = {tan_1 ( +tan’ Q +x) — Vh2+yF —x 1[tan 2+ta /h2+y(C.2)and b is the length of wall and Ii is the height of wall as shown in Figure C.1.This method, together with the dimensions of the scale model, is used to calculatefor each thermcouple location. By repeating calculations for points over the model amap of view factors can be produced.201bng the view factOtfor calcUla 11 4- SVStCd c0Or0h1I eletT anFigure C.1:‘;it p on tue groufl’1of a wall oryp(x,y)202C.2 Calculation of b3 for locations in grass parks with tree bordersThe presence of model “trees” complicates the derivation of the view factors. Spheres areused to approximate the shape of deciduous trees and the formula presented by Howell(1982) is used to determine the view factor of tree (b):= 2(1 +D2) { 1X2 — 4R2(1 + D2)J — i} (C.3)where H = x/l, R = r/l, D = y/l and X = R2 + D2 + H2 +1 with the elements x, y, 1, r,given in figure C.2. The sky view factor is then calculated:P(x,y)(C.4)Figure C.2: Definitions of elements and coordinate system for calculating the view factorof a deciduous tree for the point P.203C.3 Calculation of in savannah, garden and forest parksTo estimate for savannah, garden and forest parks, fisheye—lens photographs weretaken at specific locations across the park surface. The images were then projected ontopolar graph paper and was derived using the method of Steyn (1980).204


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items